We recently had our main gitolite server going unresponsive. It was essentially swapping heavily. The OM killer even started to kick as well. This is a VM: 1Gb memory, 2 CPUs. Of course, wait for IO was very high but on swap partition, not the repo partition. Load was 70. It was not going below 25, anyway too high. Oh, it is also sles11sp2 distro and also running gitweb. No git daemon is running. Network was sane. A reboot. A reboot fixed the issue without the load going high again. Gitolite 3.6.1 ...
Questions:
- could that be the "repack" issue causing this as explained here ? I could not tell with "ps"...
http://gitolite.com/server-sizing.html
- anyone already experienced this ? If so, probable solutions ?
- any clues ? :)
Thanks in advance