/var/log/pveproxy/access.log HUGE

[ljhardy]

I installed Pulse and it's working fine, I love the UI. However, in several hours my /var/log/pveproxy/access.log is massive and blew up my log2ram configuration, presumably logging all of the API hits coming from Pulse. Probably more of a Proxmox question, but how to limit this? access.log was over 40m in only about 8 hours.

[rcourtman]

Hey, thanks for flagging this

I just pushed an optimization to the develop branch that should fix this. Instead of making individual API calls for each VM (which was creating tons of log entries), it now uses Proxmox's bulk /cluster/resources endpoint. This reduces API calls by 95%+ while keeping the same 2-second polling.

Example: If you have 20 VMs, the old way made 20 API calls every 2 seconds. Now it makes just 1.

This will be included in the next release. Should dramatically reduce your pveproxy log growth without needing to mess with Proxmox settings or compromise Pulse's responsiveness.

Thanks again for reporting this

[ljhardy]

Does 3.31.2 include the enhancements to reduce log size?

[rcourtman]

Yes, v3.31.2 includes the bulk endpoint optimization. Changed from individual VM status calls to /cluster/resources in v3.30.0, which reduces API calls by 95%+ (from N calls to 1 call per polling cycle, where N = number of VMs).

Your pveproxy access logs should be much smaller now - instead of 50 API hits every 2 seconds for 50 VMs, you'll see just 1 bulk call plus a few for graphs.

[rcourtman]

Actually, I need to correct my previous comment. The bulk endpoint reduces some API calls, but we still need individual calls for RRD data (graphs) and current status (accurate I/O counters).

So the actual reduction is from 3N to 2N+1 API calls per cycle (where N = number of VMs). That's roughly a 33% reduction, not 95%.

Still should help with your log growth, but not as dramatically as I initially stated.

[ljhardy]

Ok, thanks, I'll fire it up again and see how the logs grow.

[ljhardy]

It's still filling up 96M of logs in less that a day, mostly in pveproxy/access.log. Do you know if there is any way to turn off these api logs in proxmox itself?

[rcourtman]

Thanks for your patience on this. Here's the quickest solution for your log growth issue:

To disable Proxmox API access logs, add this to /etc/pve/local/pveproxy.cfg:

nolog: 1

Then restart pveproxy: systemctl restart pveproxy

This will stop all API access from being logged, which should completely solve your log growth problem.

---

Technical details (for anyone interested):

Pulse uses Proxmox's bulk endpoint where possible, but still needs individual API calls for real-time I/O metrics (network/disk speeds). The bulk endpoint provides cached data that isn't fresh enough for 2-second monitoring intervals.

Alternative approaches if you want to keep API logs:

• Increase Pulse's update interval from 2 to 5+ seconds
• I could implement a "low API mode" that skips I/O monitoring entirely
• Configure logrotate to handle the larger logs

But honestly, disabling the API access logs is probably the simplest solution since they're rarely needed for debugging.

[ljhardy]

Are you sure that /etc/pve/local/pveproxy.cfg is the correct file. It didn't exist on either of my nodes, I created it with the "nolog: 1" line, restarted pveproxy and still see the log records:

- pulse-monitor@pam!pulse [30/06/2025:08:43:02 -0500] "GET /api2/json/nodes/proxmox02/qemu/111/snapshot HTTP/1.1" 200 387
- pulse-monitor@pam!pulse [30/06/2025:08:43:02 -0500] "GET /api2/json/nodes/proxmox02/lxc/109/snapshot HTTP/1.1" 200 977
- pulse-monitor@pam!pulse [30/06/2025:08:43:02 -0500] "GET /api2/json/nodes/proxmox02/lxc/104/snapshot HTTP/1.1" 200 977
- pulse-monitor@pam!pulse [30/06/2025:08:43:02 -0500] "GET /api2/json/nodes/proxmox02/qemu/105/snapshot HTTP/1.1" 200 311
- pulse-monitor@pam!pulse [30/06/2025:08:43:02 -0500] "GET /api2/json/nodes/proxmox02/lxc/102/snapshot HTTP/1.1" 200 977
- pulse-monitor@pam!pulse [30/06/2025:08:43:02 -0500] "GET /api2/json/nodes/proxmox02/qemu/107/snapshot HTTP/1.1" 200 316
- pulse-monitor@pam!pulse [30/06/2025:08:43:02 -0500] "GET /api2/json/nodes/proxmox02/lxc/106/snapshot HTTP/1.1" 200 977
...
...

There are some messages here from 2024 ( might be outdated? ) that suggest pveproxy logging to access.log is hardcoded and the only way around it is to symlink access.log to /dev/null.

proxmox forum

[ljhardy]

Anything further on this?

[rcourtman]

@ljhardy Have you tried the symlink approach mentioned in that forum post?

ln -sf /dev/null /var/log/pveproxy/access.log

Would be good to know if that actually stops the logging for you.