Jonathan Oxer
[Blog]
>> Writing a book: basics
Thu, Feb 28th 11:31pm 2008 >> Writing
My old conference pal Alec the Geek said to me a few days ago that it could be interesting to do a post on the process of writing a book: what's actually involved in going from a raw concept to grinning goofily at friends and family as you show them the finished product. There's a lot to it and I've seen remarkably little posted on the subject, particularly considering that so many of my friends and conference aquaintances have been through it at various stages, some of them multiple times. I'd have expected to see a bit more about it even just in passing. So to satisfy Alec's curiosity I'll have a go. It's a pretty big subject though so it'll probably end up as a series of posts. So let's start with the different publishing processes. You might think this would be the very last thing to think about, but it determines the entire process of writing from beginning to end so you need to be aware of it right from the start. I've been through the book-writing process three different ways (DIY self-published, trade published, and assisted-self-published) and they're all totally different. Of course everyone who thinks about writing a book wants it to be trade-published by a big name company: there's an incredible feeling of validation when a publisher says they want to publish your work. But getting a publisher to take you on is also incredibly hard if you don't already have a track record, and therein lies the rub. You can't get a publisher without a track record, and you can't get a track record without getting your work out there somehow. Think about the career path of a successful musician. They don't just sit around in their bedroom practicing in private and then one day out of the blue start calling major record labels looking for a deal, expecting to be signed up for a world tour playing to massive stadiums on their debut and receiving a huge signing bonus. But that's exactly what a lot of would-be authors expect. Publishers have to shift a lot of copies on quite small margins for a title to be financially viable, and as a result they don't like to gamble on unknowns. So be realistic about what you want to achieve and what your reasons are for wanting to write a book. Not all musicians get fat contracts and play to packed stadiums and have their cardboard cutout displayed in the foyer at every Sanity store, and not all writers have a publishing deal that gets their book displayed cover-out in every bookshop and airport lounge on the planet and discussed on Oprah's Book Club. You have to work up to that sort of thing and it can take a really long time or (far more likely) never happen at all. That doesn't mean it isn't worthwhile being an unpublished musician or writer. You just have to treat it as a journey, not a single step from beginner to expert. What do indie musos do when they're trying to get their name out and build a fanbase? Play local gigs and record demo tapes (or CDs, or MP3s) and practice their craft. For a writer it's exactly the same. You have to submit stories to whoever will print them, give away your work and develop a reputation. I did that for years, submitting articles to all sorts of places: one of the biggest boosts I got early on was regularly submitting stories for linmagau, a short-lived online Linux magazine created by Kimberly Shelt in Perth. I was fortunate enough to provide a couple of stories that more due to timeliness than my writing ability brought in an incredible number of readers, which made Kim extremely receptive to anything else I wanted to submit. I jumped on the opportunity and submitted articles faster than she could publish them, and then leveraged my way into other publications from there. And like a muso recording their first tracks in a home studio then burning CDs and giving them away at parties or selling them at local gigs, you can produce, print and give away or sell your early work directly yourself. Yes, that hated term with all the stigma attached: self-publishing. This post is way too long already so I'll continue with a follow-up post in a day or two with the differences between DIY self-publishing, trade publishing, and assisted self-publishing.
My old conference pal Alec the Geek said to me a few days ago that it could be interesting to do a post on the process of writing a book: what's actually involved in going from a raw concept to grinning goofily at friends and family as you show them the finished product. There's a lot to it and I've seen remarkably little posted on the subject, particularly considering that so many of my friends and conference aquaintances have been through it at various stages, some of them multiple times. I'd have expected to see a bit more about it even just in passing. So to satisfy Alec's curiosity I'll have a go. It's a pretty big subject though so it'll probably end up as a series of posts. So let's start with the different publishing processes. You might think this would be the very last thing to think about, but it determines the entire process of writing from beginning to end so you need to be aware of it right from the start. I've been through the book-writing process three different ways (DIY self-published, trade published, and assisted-self-published) and they're all totally different. Of course everyone who thinks about writing a book wants it to be trade-published by a big name company: there's an incredible feeling of validation when a publisher says they want to publish your work. But getting a publisher to take you on is also incredibly hard if you don't already have a track record, and therein lies the rub. You can't get a publisher without a track record, and you can't get a track record without getting your work out there somehow. Think about the career path of a successful musician. They don't just sit around in their bedroom practicing in private and then one day out of the blue start calling major record labels looking for a deal, expecting to be signed up for a world tour playing to massive stadiums on their debut and receiving a huge signing bonus. But that's exactly what a lot of would-be authors expect. Publishers have to shift a lot of copies on quite small margins for a title to be financially viable, and as a result they don't like to gamble on unknowns. So be realistic about what you want to achieve and what your reasons are for wanting to write a book. Not all musicians get fat contracts and play to packed stadiums and have their cardboard cutout displayed in the foyer at every Sanity store, and not all writers have a publishing deal that gets their book displayed cover-out in every bookshop and airport lounge on the planet and discussed on Oprah's Book Club. You have to work up to that sort of thing and it can take a really long time or (far more likely) never happen at all. That doesn't mean it isn't worthwhile being an unpublished musician or writer. You just have to treat it as a journey, not a single step from beginner to expert. What do indie musos do when they're trying to get their name out and build a fanbase? Play local gigs and record demo tapes (or CDs, or MP3s) and practice their craft. For a writer it's exactly the same. You have to submit stories to whoever will print them, give away your work and develop a reputation. I did that for years, submitting articles to all sorts of places: one of the biggest boosts I got early on was regularly submitting stories for linmagau, a short-lived online Linux magazine created by Kimberly Shelt in Perth. I was fortunate enough to provide a couple of stories that more due to timeliness than my writing ability brought in an incredible number of readers, which made Kim extremely receptive to anything else I wanted to submit. I jumped on the opportunity and submitted articles faster than she could publish them, and then leveraged my way into other publications from there. And like a muso recording their first tracks in a home studio then burning CDs and giving them away at parties or selling them at local gigs, you can produce, print and give away or sell your early work directly yourself. Yes, that hated term with all the stigma attached: self-publishing. This post is way too long already so I'll continue with a follow-up post in a day or two with the differences between DIY self-publishing, trade publishing, and assisted self-publishing.
>> Rackspace Shmackspace
Wed, Feb 27th 10:09am 2008 >> Bad People
Don't you hate it when what appears to be a good reputation turns out to actually be just overpowered marketing? We're running upwards of 80 servers here in Melbourne and growing all the time, but we need to start rolling out some infrastructure in the US as well. I've used a few US-based dedicated server providers in the past including Servepath, Dreamhost, and others, and had generally good experiences with them. This time though I wanted to start with a machine that we could do virtualisation on and start small, with just a few VMs initially and then provision more physical and virtual hosts as needed, and being so far away I wanted to be sure any problems we had would be fixed ASAP since we don't have physical access to the boxes. And what I really wanted was either an Ubuntu or Debian host so it would be consistent with our server SOE and our purpose-built cluster management tools would work nicely with it. In the end I decided to go with Rackspace on the strength of their reputation for tech support: 24 hour phone hotline, dedicated account manager, etc. They couldn't provide an Ubuntu or Debian machine, but said they could provide RHEL which has good Xen support. That sucked a bit because it means we can't integrate the machine into our management infrastructure, but I was willing to put up with that annoyance (and pay more than twice what I would have elsewhere) to get the "fanatical support" that they promise. A trade-off, but I thought it was worthwhile. The problems began right from the start. I don't think I've ever asked a single question of their tech support and got a straight answer the first time. It's usually been a case of asking the question, getting a useless answer, saying "no, that's not what I wanted to know, the original question was...", getting another obtuse answer, repeat until enough information could be gleaned. In all my dealings with them it's been as if they simply don't hear what I say, and I have to keep repeating myself until they do. A classic example of missing the point was early on when they supplied the server with RHEL4 after I had spent much time on the phone explaining exactly what/why we needed the server: primarily as a virtualisation platform. But RHEL4 doesn't have good Xen support, so it wasn't any use to us! So they suggested reinstalling the server with RHEL5 in order to provide all the necessary Xen packages, which they then did. After it had been reinstalled I logged in and tried to get the Xen tools installed and found they weren't available through Yum, and I couldn't even install a Xen-capable kernel. After hassling tech support a bit about why there were no Xen packages available, they responded saying that when the server was reinstalled they'd forgotten to enable the virtualisation channel. Hello? Is anyone listening? That was a trivial problem compared with the network shenannigans. I had asked for additional IP addresses so we could use them for virtual machines, and the entries in the trouble ticket go something like this (paraphrased for brevity): Me: "I've been provided one additional IP address but no details of netmask, gateway, etc. I have brought up a dummy interface to test it but I need to know what the details should be". Them: "Here are the details of your interface: " (they then pasted in the 'ifconfig' output from the dummy interface that *I* had created!) Me: "No, that's the one I created and it didn't answer my question about what the netmask and gateway should be." (Then while I'm logged into the server eth0 mysteriously vanishes, and I'm cut off) Me: "Eth0 has gone away." Them: "Sorry, we dropped it by mistake. Here are the additional network details:" (they provided a single address with a .248 netmask, with the address not at the start of the range but at the second available host address and no further explanation). Me: "This server was set up for virtualisation. The additional address provided is on a different subnet to the host address and can't be bridged to the host adaptor." (I experiment a bit to figure out what's going on with the routing of the subnet they've specified) Me: "It looks like the entire /29 is being routed to our server, but we've been told we have a single additional IP address available. What about the rest of the range?" (I experiment a bit more, adding one of the other IP addresses in the range as an alias to the host adaptor, set up a virtual bridge, set up a virtual machine, set the route on the VM through the aliased IP on the host, and check that it works. It does, but at this point I've effectively hijacked an IP address that I hadn't been authorised to use). Me: I explain what I've done above, and ask *again* for clarification on what additional IP's we've actually been assigned and the network configuration. Them: Reply saying that yes, my analysis was correct and in fact the entire /29 is routed to our server after all. Elapsed from first request to actual helpful answer (which was really just saying that what I had figured out for myself was correct): 20 hours, and 10 comments in the trouble ticket. Grrr, why couldn't they have just answered the bloody question in the first place? Those weren't the only problems I had, but they're indicative of my whole Rackspace experience. Maybe I was just having a bad run. Maybe because our requirements are a bit outside the norm they're having trouble providing us with support. That's understandable, so let's part as friends and terminate the contract, no major harm done. So I went looking for a cancellation form on their site or in their support area. No dice. Then I submitted a support ticket asking about the cancellation procedure. Not only did the response not answer my actual question (as usual), but I got a hard-ass response saying that we were locked into a contract and they were going to hold us to it until the day it expires. *That* blew my top. I'm not giving them the benefit of the doubt anymore: now I'm just pissed off. Any company that actually cared about customer service would understand that the small financial gain from holding a disgruntled customer to the exact terms of their contract are far outweighed by the resentment and bad will that it causes. Until that response to my cancellation request I would have walked away disappointed that it hadn't worked out, but not upset. Now I'm mad. Then to really rub salt in the wound, the response also said: "With specific regards to your configuration, the Managed Department of Rackspace does not support Virtualization at all, and I believe that the majority of your issues with our Support Team originate with this fact." They don't support virtualisation at all? Hmmm. I've written enough already, so I'll end with this screencap from the Rackspace homepage. Pay close attention to the message across the top, then the bit in the bottom right corner.
Fanatical support indeed.
Don't you hate it when what appears to be a good reputation turns out to actually be just overpowered marketing? We're running upwards of 80 servers here in Melbourne and growing all the time, but we need to start rolling out some infrastructure in the US as well. I've used a few US-based dedicated server providers in the past including Servepath, Dreamhost, and others, and had generally good experiences with them. This time though I wanted to start with a machine that we could do virtualisation on and start small, with just a few VMs initially and then provision more physical and virtual hosts as needed, and being so far away I wanted to be sure any problems we had would be fixed ASAP since we don't have physical access to the boxes. And what I really wanted was either an Ubuntu or Debian host so it would be consistent with our server SOE and our purpose-built cluster management tools would work nicely with it. In the end I decided to go with Rackspace on the strength of their reputation for tech support: 24 hour phone hotline, dedicated account manager, etc. They couldn't provide an Ubuntu or Debian machine, but said they could provide RHEL which has good Xen support. That sucked a bit because it means we can't integrate the machine into our management infrastructure, but I was willing to put up with that annoyance (and pay more than twice what I would have elsewhere) to get the "fanatical support" that they promise. A trade-off, but I thought it was worthwhile. The problems began right from the start. I don't think I've ever asked a single question of their tech support and got a straight answer the first time. It's usually been a case of asking the question, getting a useless answer, saying "no, that's not what I wanted to know, the original question was...", getting another obtuse answer, repeat until enough information could be gleaned. In all my dealings with them it's been as if they simply don't hear what I say, and I have to keep repeating myself until they do. A classic example of missing the point was early on when they supplied the server with RHEL4 after I had spent much time on the phone explaining exactly what/why we needed the server: primarily as a virtualisation platform. But RHEL4 doesn't have good Xen support, so it wasn't any use to us! So they suggested reinstalling the server with RHEL5 in order to provide all the necessary Xen packages, which they then did. After it had been reinstalled I logged in and tried to get the Xen tools installed and found they weren't available through Yum, and I couldn't even install a Xen-capable kernel. After hassling tech support a bit about why there were no Xen packages available, they responded saying that when the server was reinstalled they'd forgotten to enable the virtualisation channel. Hello? Is anyone listening? That was a trivial problem compared with the network shenannigans. I had asked for additional IP addresses so we could use them for virtual machines, and the entries in the trouble ticket go something like this (paraphrased for brevity): Me: "I've been provided one additional IP address but no details of netmask, gateway, etc. I have brought up a dummy interface to test it but I need to know what the details should be". Them: "Here are the details of your interface: " (they then pasted in the 'ifconfig' output from the dummy interface that *I* had created!) Me: "No, that's the one I created and it didn't answer my question about what the netmask and gateway should be." (Then while I'm logged into the server eth0 mysteriously vanishes, and I'm cut off) Me: "Eth0 has gone away." Them: "Sorry, we dropped it by mistake. Here are the additional network details:" (they provided a single address with a .248 netmask, with the address not at the start of the range but at the second available host address and no further explanation). Me: "This server was set up for virtualisation. The additional address provided is on a different subnet to the host address and can't be bridged to the host adaptor." (I experiment a bit to figure out what's going on with the routing of the subnet they've specified) Me: "It looks like the entire /29 is being routed to our server, but we've been told we have a single additional IP address available. What about the rest of the range?" (I experiment a bit more, adding one of the other IP addresses in the range as an alias to the host adaptor, set up a virtual bridge, set up a virtual machine, set the route on the VM through the aliased IP on the host, and check that it works. It does, but at this point I've effectively hijacked an IP address that I hadn't been authorised to use). Me: I explain what I've done above, and ask *again* for clarification on what additional IP's we've actually been assigned and the network configuration. Them: Reply saying that yes, my analysis was correct and in fact the entire /29 is routed to our server after all. Elapsed from first request to actual helpful answer (which was really just saying that what I had figured out for myself was correct): 20 hours, and 10 comments in the trouble ticket. Grrr, why couldn't they have just answered the bloody question in the first place? Those weren't the only problems I had, but they're indicative of my whole Rackspace experience. Maybe I was just having a bad run. Maybe because our requirements are a bit outside the norm they're having trouble providing us with support. That's understandable, so let's part as friends and terminate the contract, no major harm done. So I went looking for a cancellation form on their site or in their support area. No dice. Then I submitted a support ticket asking about the cancellation procedure. Not only did the response not answer my actual question (as usual), but I got a hard-ass response saying that we were locked into a contract and they were going to hold us to it until the day it expires. *That* blew my top. I'm not giving them the benefit of the doubt anymore: now I'm just pissed off. Any company that actually cared about customer service would understand that the small financial gain from holding a disgruntled customer to the exact terms of their contract are far outweighed by the resentment and bad will that it causes. Until that response to my cancellation request I would have walked away disappointed that it hadn't worked out, but not upset. Now I'm mad. Then to really rub salt in the wound, the response also said: "With specific regards to your configuration, the Managed Department of Rackspace does not support Virtualization at all, and I believe that the majority of your issues with our Support Team originate with this fact." They don't support virtualisation at all? Hmmm. I've written enough already, so I'll end with this screencap from the Rackspace homepage. Pay close attention to the message across the top, then the bit in the bottom right corner.
Fanatical support indeed.>> Quickstart Guide to Google AdWords
Mon, Feb 25th 2:12pm 2008 >> Web Development
At CeBIT last year I grabbed a bunch of little booklets called "Insiders Guide to Google AdWords" from the Google stand, and since then we've been giving them away to interested customers. Then about a month ago in eBusiness News I did a story on the Google AdWords bid optimizer and offered to send readers the last few copies I had of Insiders Guide.
The response was unbelievable. My last few copies were claimed within five minutes of eBusiness News going out, and I ended up with a spreadsheet of people who wanted it but missed out.
"No problem," I thought, "I'll just get on the phone to the AdWords marketing team and grab some more copies. A box or two should do it."
No such luck. It turns out that Insiders Guide is out of print, and the AdWords marketing team has no intention of producing more - which is a pity, because it's a useful little guide. Oh well, since I talk to a lot of people about online marketing and many of them want to try out AdWords, perhaps I could arrange to have extra copies printed at my expense if Google didn't have plans to produce more?
"Absolutely not. It's Google copyright material and nobody else may reproduce it under any circumstances."
Wow. So they don't want me helping to advertise their services and send them more customers by reproducing what is effectively their marketing material for them at my expense? OK, I can understand them wanting to protect their corporate image, but still!
That left me in a pickle. Many of my customers are hassling me for a quickstart guide to AdWords, but Google won't let me use theirs. What's a guy to do?
Write one myself, of course!
So over the space of a couple of weeks I spent many late evenings writing "Quickstart Guide to Google AdWords", and now I'm happy to say it's all done so maybe I can start getting some sleep again. The first hard copies should arrive in Oz late this week or early next week but you can already order it online, and it should appear on Amazon.com and B&N and other random places once the Books In Print database does it's slow-as-a-snail update thing.
At CeBIT last year I grabbed a bunch of little booklets called "Insiders Guide to Google AdWords" from the Google stand, and since then we've been giving them away to interested customers. Then about a month ago in eBusiness News I did a story on the Google AdWords bid optimizer and offered to send readers the last few copies I had of Insiders Guide.
The response was unbelievable. My last few copies were claimed within five minutes of eBusiness News going out, and I ended up with a spreadsheet of people who wanted it but missed out.
"No problem," I thought, "I'll just get on the phone to the AdWords marketing team and grab some more copies. A box or two should do it."
No such luck. It turns out that Insiders Guide is out of print, and the AdWords marketing team has no intention of producing more - which is a pity, because it's a useful little guide. Oh well, since I talk to a lot of people about online marketing and many of them want to try out AdWords, perhaps I could arrange to have extra copies printed at my expense if Google didn't have plans to produce more?
"Absolutely not. It's Google copyright material and nobody else may reproduce it under any circumstances."
Wow. So they don't want me helping to advertise their services and send them more customers by reproducing what is effectively their marketing material for them at my expense? OK, I can understand them wanting to protect their corporate image, but still!
That left me in a pickle. Many of my customers are hassling me for a quickstart guide to AdWords, but Google won't let me use theirs. What's a guy to do?
Write one myself, of course!
So over the space of a couple of weeks I spent many late evenings writing "Quickstart Guide to Google AdWords", and now I'm happy to say it's all done so maybe I can start getting some sleep again. The first hard copies should arrive in Oz late this week or early next week but you can already order it online, and it should appear on Amazon.com and B&N and other random places once the Books In Print database does it's slow-as-a-snail update thing.>> FOSS GeoSpatial conference coming to Sydney
Mon, Feb 18th 11:05am 2008 >> Misc
One of my last official actions at Linux Australia was to write a letter of support for the team bidding to bring the Free & Open Source Software for GeoSpatial (FOSS4G) conference to Sydney, and a few minutes ago I got an email letting me know that their bid had been successful. So in November 2009 we're going to have one of the best geospatial conferences in the world right on our doorstep: wiki.osgeo.org/index.php/FOSS4G_2009_Press_Release_1 Cool!
One of my last official actions at Linux Australia was to write a letter of support for the team bidding to bring the Free & Open Source Software for GeoSpatial (FOSS4G) conference to Sydney, and a few minutes ago I got an email letting me know that their bid had been successful. So in November 2009 we're going to have one of the best geospatial conferences in the world right on our doorstep: wiki.osgeo.org/index.php/FOSS4G_2009_Press_Release_1 Cool!
>> I welcome our Slashdot overlords
Fri, Feb 15th 1:56pm 2008 >> Tech Toys
Last night just before I turned off the screen on my MythTV box to go to bed I pulled up the newsfeeds, and smack at the top of the Slashdot headlines was a story titled "Australia's Geekiest Man". My immediate thought was "that should have been me - I want to know who around here is even geekier than I am!" so I clicked through to see who it was. Sigh. It was me. I don't know whether to be honored or offended about the label! AJ has already called me "Australia's manliest geek", but *that's* taking it just a bit too far ;-) The story seems to have started a bit of a circus. I've already had about 6 journos contact me for follow-up stories, I've had 3 consulting job offers relating to RFID and automation just this morning, and two companies have offered to send me automation equipment to test out (no doubt so I can blog about it and get them more sales). Now I just need "Australia's Geekiest Man" business cards.
Last night just before I turned off the screen on my MythTV box to go to bed I pulled up the newsfeeds, and smack at the top of the Slashdot headlines was a story titled "Australia's Geekiest Man". My immediate thought was "that should have been me - I want to know who around here is even geekier than I am!" so I clicked through to see who it was. Sigh. It was me. I don't know whether to be honored or offended about the label! AJ has already called me "Australia's manliest geek", but *that's* taking it just a bit too far ;-) The story seems to have started a bit of a circus. I've already had about 6 journos contact me for follow-up stories, I've had 3 consulting job offers relating to RFID and automation just this morning, and two companies have offered to send me automation equipment to test out (no doubt so I can blog about it and get them more sales). Now I just need "Australia's Geekiest Man" business cards.
>> Virtual buttons, real appliance
Sun, Feb 10th 9:07am 2008 >> DIY
Following on from my tutorial at linux.conf.au on hardware / software hacking to join Second Life to the real world, I've now put up the first of a series of step-by-step instructions. This one is to create a couple of virtual buttons inside Second Life that control an actual physical appliance in the real world when they're clicked: Virtual Buttons Real Appliance
I still need to add a bit of stuff at the end about connecting relays and modifying an appliance remote control, but most of it's there.
Following on from my tutorial at linux.conf.au on hardware / software hacking to join Second Life to the real world, I've now put up the first of a series of step-by-step instructions. This one is to create a couple of virtual buttons inside Second Life that control an actual physical appliance in the real world when they're clicked: Virtual Buttons Real Appliance
I still need to add a bit of stuff at the end about connecting relays and modifying an appliance remote control, but most of it's there.>> Office chaos
Fri, Feb 8th 2:39pm 2008 >> IVT
We're currently part way through a massive office refit at Internet Vision Technologies, partly to improve the general work environment and partly to make better use of space since we moved all our hosting infrastructure offsite. As a result it's a disaster area at the moment. We've shoved *everything* into the back half of the building while the front is refitted, then we'll shift everything forward so the back half can be done. So I'm currently jammed into a corner under a bundle of ethernet and just working off my laptop:
Likewise Antoine and Steve have desks that look like they're working in a war zone:
And this is Andrew having a chat to Senior, with Luke's temporary desk just to the right:
Matt probably has the best setup of all: he claimed the old boardroom table and set himself up on it near Neil and Dianne's desks:
This morning a couple of the guys had a meeting with a client so they quickly assembled a temporary "meeting room" in the middle of the area where all the new desks will be assembled:
Just behind that the table tennis table has also been set up temporarily, since it's been moved out of the back room to make space for desks and random junk:
Looking into the new meeting room (with the larger boardroom beyond it) shows how it's starting to come together. Some time next week there will be a glass wall extending right along here to separate the meeting room and boardroom from the work area:
So it's a total mess right now, but in a couple of weeks we'll have a shiny new office with that new-car smell :-)
We're currently part way through a massive office refit at Internet Vision Technologies, partly to improve the general work environment and partly to make better use of space since we moved all our hosting infrastructure offsite. As a result it's a disaster area at the moment. We've shoved *everything* into the back half of the building while the front is refitted, then we'll shift everything forward so the back half can be done. So I'm currently jammed into a corner under a bundle of ethernet and just working off my laptop:
Likewise Antoine and Steve have desks that look like they're working in a war zone:
And this is Andrew having a chat to Senior, with Luke's temporary desk just to the right:
Matt probably has the best setup of all: he claimed the old boardroom table and set himself up on it near Neil and Dianne's desks:
This morning a couple of the guys had a meeting with a client so they quickly assembled a temporary "meeting room" in the middle of the area where all the new desks will be assembled:
Just behind that the table tennis table has also been set up temporarily, since it's been moved out of the back room to make space for desks and random junk:
Looking into the new meeting room (with the larger boardroom beyond it) shows how it's starting to come together. Some time next week there will be a glass wall extending right along here to separate the meeting room and boardroom from the work area:
So it's a total mess right now, but in a couple of weeks we'll have a shiny new office with that new-car smell :-)>> Who ate my disk?
Thu, Feb 7th 9:08am 2008 >> Linux
I'm confuzzled. Something is eating up disk on one of my VMs, and I can't figure out what. On a machine that isn't even in active use because it's only accessed every week or so for demos, and which should have way less than 50% disk utilisation:
I'm confuzzled. Something is eating up disk on one of my VMs, and I can't figure out what. On a machine that isn't even in active use because it's only accessed every week or so for demos, and which should have way less than 50% disk utilisation:
root@intra-174:/# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/hda2 20G 18G 1.6G 92% /
varrun 193M 176K 192M 1% /var/run
varlock 193M 4.0K 193M 1% /var/lock
udev 193M 24K 193M 1% /dev
devshm 193M 0 193M 0% /dev/shm
Getting tight. So let's see where all that disk is being used:
root@intra-174:/# du -h --max-depth=1
16K ./lost+found
3.4M ./bin
33M ./boot
4.0K ./cdrom
3.0M ./dev
36M ./etc
4.0K ./floppy
260K ./home
4.0K ./initrd
168M ./jade
262M ./lib
4.0K ./media
4.0K ./mnt
4.0K ./opt
395M ./proc
226M ./root
9.8M ./sbin
483M ./sitebuilder
481M ./srv
0 ./sys
202M ./tmp
1.9G ./usr
1.5G ./var
5.6G .
Umm, what the hell? 5.6G != 18G.
At linux.conf.au I had a quick chat to Tony Breeds and Steve Hanley about it, who both speculated that the problem could be sparse file preallocation being reported by du. So let's have a look:
root@intra-174:/# du -h --apparent-size --max-depth=1
16K ./lost+found
3.2M ./bin
32M ./boot
4.0K ./cdrom
244K ./dev
29M ./etc
4.0K ./floppy
144K ./home
4.0K ./initrd
168M ./jade
241M ./lib
4.0K ./media
4.0K ./mnt
4.0K ./opt
393M ./proc
226M ./root
9.4M ./sbin
482M ./sitebuilder
273M ./srv
13M ./sys
199M ./tmp
1.7G ./usr
1.5G ./var
5.1G .
The odd thing is that rebooting this VM drops the reported disk usage in df back down under 6G, then it starts creeping back up again at about a gig per day.
Any ideas?
1 of 1
[ Back to top ]
