OK. I know what this is. It is an OOM (Out Of Memory) issue in your Virtuozzo VPS.qlex wrote:QlexCode: Select all
Downloading packages Running rpm_check_debug Running Transaction Test Killed
Bottom line: You have centos 5.x (where x < 6) installed at the moment, and when you run yum update things go wrong because yum uses too much memory when doing so as there's a heck of a lot for yum to do and figure out. Unfortunately this causes Virtuozzo to kill the yum process and leave you with incomplete transactions. BUT, worst of all, when you try to do run yum update again, yum will simply stall, unkillably, invisibly. It does not help that you are trying to update Atomic stuff (e.g. php) at the same time.
Here's my suggestion. No guarantees. At your own risk! I'd let your provider do whatever he things best personally, as he'll be able to see what's happening and deal with it as required. But if you decide not to, here's what might help:
1) Reboot your VPS
2) Install yum-utils [yum install yum-utils](You will see an error when you do anything with yum following your earlier "killed" issue, which talks about incomplete transaction. Do not worry. We will deal with this shortly, and yum-utils is what we'll use to deal with it.
2a) I'd disable the Atomic repo in order not to upgrade php to 5.3 right now. Update the OS THEN update php/mysql etc.
3) Follow the Centos-recommended method to update from an ealier version of Centos Centos (5.4) to Centos 5.6. If you already have 5.5 it isn't strictly necessary to do this, but it will help you anyway.
This is as follows:
4) Now, instead of running yum update right at the end there, reboot your VPS THEN run yum update. This should allow yum to do its thing with plenty of memory.yum clean all
yum update glibc\*
yum update yum\* rpm\* pyth\*
yum clean all
yum update mkinitrd nash
yum update (do not do this yet though!)
5) Re-enable the Atomic repo.
6) run yum-complete-transactions
This will clear your previous errors. It may well also install the atomic php and other stuff because they will be incomplete from the first time you ran yum update, and yum-complete-transactions will resolve that.
HOWEVER, I'm not completely sure what will happen with the Atomic repo disabled in step 2a. In may be that doing that is a bad idea and you should leave it enabled. I'm suggesting that you disable it in order to reduce the potential for problems, but there's a possibility that it will cause more problems with it disabled than with it enabled. I'm sorry but I don't know for sure -- maybe someone else will comment on this?
Incidentally, I know it is an OOM because it has happened on three Centos 5 VPS I've updated from 5.5 to 5.6 under Virtuozzo, and I know this is a VPS and I'm reasonably sure it is a Virtuozzo one. The method I've outlined (rebooting before final yum update) worked 100% of the time on the other VPSes I updated once I'd figured out what was going on. You will not find anything in /var/log/messages in the contrainer, but your hosting company will see the OOM errors in /var/log/messages in the hardware node itself.
Faris.
