I have a site running on several Linode virtual machines, including separate VMs running:
- Django (and apache)
- Django (for running memory-intensive scripts)
- Postgres
- Media server, with a private network mount point on the Django VMs.
It has suddenly started developing two issues at once:
- The first is that postgres complains of having 'sorry, too many clients already'.
- The second is that the media server is no longer accessible from the django sites. If I try to ping the media server's local IP, I get a 'Destination unreachable'.
How to solve it (short term)
Powering off the second Django machine (the one for memory-intensive tasks) and does not solve the issue.
The problem is resolved by restarting the postgres and media boxes, but it tends to recur after a few days.
Other info
Mostly, but not always, both issues seem to come along at the same time. I have not been able to find a pattern to when they happen, sometimes at busy periods, sometimes not, though it does seem to be more likely when an intensive task is running.
The site has been running without this issue for a few years, and has only just started happening.
I have looked at the /var/log/syslog for the different boxes but nothing looks very helpful.
Here is /etc/fstab for the mount of the media VM on the Django server (anonymised):
192.168.xxx.xxx:/path/to/mount /mnt/media nfs rsize=8192,wsize=8192,timeo=14,intr
I am unsure where to look next in order to solve this - does anyone have an idea? Thank you in advance!
