Recently added a few H@H client slots so I decided to test with lower-end boxes to see if they can work. And unfortunately there were problems on 2 of them.
Common characteristics are: (1) OpenVZ; (2) 256MB RAM.
Why I think these 2 should be the factor:
1) I have other OpenVZ boxes with 512MB RAM running without problems.
2) I have at least 1 Xen box with 256MB RAM running without problems.
Common problems:
1) Could eventually hang when running, or even hang on startup when loading db; when hanged, it cannot be killed normally, and need to be killed using "kill -9"
2) Even it does not hang, it could eventually terminate after a few hours with a dump like this:
hs_err_pid521.log.txt ( 16.6k )
Number of downloads: 20Tried these methods to tackle the problem but no luck:
1) Specifying all, any or none of these options (for the H@H jar): --disable_logging, --disable_bwm, --use_more_memory
2) Specifying like "-Xmx160m" to limit its use of memory, but it seems the process dies even faster
3) Tried different OS, e.g. CentOS 6, Debian 6, Ubuntu, but same symptoms happened basically
What's more, I have upgraded 1 of the problematic boxes to a higher package with 512MB RAM and the problem got solved eventually. For the other one, I am still working on that as I have made a longer subscription plan, and want to see if there could be any solution other than upgrading.
Any suggestion would be appreciated. Thanks.
This post has been edited by mnching: Jan 20 2013, 21:10