-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 If you can write to the hard drive, you can make a swapfile:
Make a 512MB swapfile dd if=/dev/zero of=/vps.swp bs=1M count=512 Make it swap mkswap /vps.swp Mount it swapon /vps.swp Verify it is being used swapon -s That HAS to work...add it to the fstab and you're cooking with grease If this doesn't work then kill me Bryan Chris Miller wrote: | Bryan Smith wrote: |> Chris, |> |> Everyone gave great answers, but why reinvent the wheel when there are |> awesome packages that will do exactly what you want with a web-interface |> as well. Enter monit(apt-get install monit)...it does exactly what is |> sounds like it does; monitor daemons. |> |> Monit will actually attempt to connect to your daemons at a set |> interval(1 minute on my system) and lie dormant if there is no alarm. |> When monit fails to connect it sends an email that the process is down |> and restarts it for you...it also send an email letting you know that |> the process is back up. It does ssl as well...this is what you need, it |> is a script indeed but has a bit more intelligence than the run of the |> mill cronjob. I use it for snmp, ldap, nscd, apache, postfix, blah and |> blah blah. This thing is totally customizable and there are tons of |> examples out in the wild. | | Sounds great. I'll look into it. | |> There is one file to config and it is super easy at that. A LOT of |> people use monit to keep mongrels stuff rolling. We are not really |> addressing the problem though Chris...how can you just run out of |> memory? No swap partition? Swap not big enough? Did you make a swap file |> even? How do you know these things are crashing because they are out or |> memory? What error does apache throw to you? | | It's literally the system running out of memory. It's a $15/month VPS, | so memory is little. I get 317 MB for RAM and no swap space (I tried to | create a file, mkswap and swapon - no luck... makes sense, how would | they make their money otherwise?) I'm pretty certain they're out of | memory because I'm always running within 30 MB of the total main memory | limit. All that has to happen is for one process to spike, to not | garbage collect soon enough, etc. and then the system will run genuinely | and completely out of memory. The kernel will try and find a scapegoat | to butcher at the High Tribunal of Memory Abusers, and it usually ends | up being one of my more important processes. I haven't really got a | problem if the server fails to deliver a page every once in a while, but | it needs to pick itself back up (immediately if at all possible) to make | the user happy when he/she hits "refresh." | |> Bryan |> |> Chris Miller wrote: |> | Robert Citek wrote: |> |> On Sat, Sep 6, 2008 at 2:34 AM, Chris Miller |> |> <[EMAIL PROTECTED]> wrote: |> |>> I was wondering if anyone knew of a tool that could provide a bit more |> |>> immediate results, constantly verifying that a process or daemon is |> |>> running, and to start it if it's not. I'd settle for something that |> |>> runs every minute (that isn't a cron job, since they leave me feeling |> |>> very icky, kind of like a lame hack. It works, but there should be a |> |>> better solution). |> |> Most system processes (e.g. those in /etc/init.d/) create a .pid file |> |> in /var/run: |> |> |> |> $ ls -la /var/run/*.pid |> |> |> |> You can use that as a way to check if a system is up and running. |> |> Using apache as an example: |> |> |> |> ps -p $(cat /var/run/apache2.pid 2> /dev/null ) >& /dev/null || |> |> echo "restart apache" |> |> |> |> Why the beef with crontab? |> | |> | It leaves me with the icky feeling of a bad hack. I keep feeling that |> | there should be a more elegant, problem-specific, fine-tuned solution. |> | Something written in a real programming language like C with cryptic |> | configuration file documentation and near ubiquitous, yet hilariously |> | unspoken use all over the place. |> | |> | Yes, I am aware that I just described crontab. |> | |> |>> I was thinking that there's bound to be a kind of "process nazi" tool |> |>> that'll keep things running smoothly, but I don't know of any. I'm |> |>> running Debian 5. |> |> I'm guessing there's a reason you are running Debian 5 despite it |> |> being labeled as "testing": |> | |> | Heh. You've probably never tried to set up Ruby on Rails on Debian. |> | Let's just say that the maintainers weren't aware that there was a |> | hideous flaw in the supplied software, and I'm too lazy to downgrade to |> | a supposedly more stable version after having done the work to upgrade. |> | |> | The tale of joys and woes to get my site working is here: |> | |> | http://lordsauron.wordpress.com/2008/06/18/zero-to-redmine-in-22-steps/ |> | |> | Please notice that I disclose all my configuration secrets not because |> | I'm confident my server is super-secure (I'm not) but because it was a |> | PITA to get running, and I don't want anyone to ever have to beat |> | through all that nonsense without documentation again. |> | |> | | | | | - -- A healthy diet includes Linux, Linux and more Linux. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFIwue9h+MLjl5SKYQRAkSRAJ0Tw9qFbzOyzQQOJ/jLMgSPbuZ70QCfVl9R ST71n3rYGu/PgbN4aEfWU8k= =9tLU -----END PGP SIGNATURE----- --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Linux Users Group. To post a message, send email to [email protected] To unsubscribe, send email to [EMAIL PROTECTED] For more options, visit our group at http://groups.google.com/group/linuxusersgroup -~----------~----~----~----~------~----~------~--~---
