Switch memory leaks and reboots

[dshef]

We have several switches that are slowly using more and more RAM, and then they finally reboot. Some reboot after a couple of weeks, some reboot after a couple of days, and the only culprit I can find is the memory usage goes to ~100% and then I know it will reboot soon. We use Mikrotik "the dude", and librenms for monitoring these switches. Unimus also backs up once a day, using SSH. What can we do to fix this problem? I appreciate your help!

edit: we have models WS-12-250-DC, WS-12-DC, WS-24-400A, WS-26-400-AC, WS-26-400-IDC
we have a bit of version disparity, but use 1.5.2, 1.5.5, 1.5.6, and 1.5.8

Dave

[dshef]

1.5.2: one unit reboots every 14 days on average (sees about 120Mbps max), the other has 142 days uptime (sees about 300Mbps max)
1.5.5: one unit, uptime 119days
1.5.6 one unit, uptime 54days
1.5.8 average uptimes: 5.83days, 10.11days, 9.67days, 8.16days, 6.98days (and there are 5 more that don't seem to be a problem)

[sakita]

The older firmware did have a memory leak problem with the eventual reboot because of it.

We have over 100 switches on 1.5.8 and aren't experiencing these issues (as of the date of this post the only newer version for the WS line is 1.5.9 rc1).

[dshef]

most of my problems are on 1.5.8. please tell me I don't need to factory reset or something.

would this problem be caused by poor storm control configuration?

[Stephen]

dshef,

That is a possibility. Factory resetting is something that you could try, but obviously that is a complicated option.
Something other options you could try would be: disabling all Discovery options if you have any active, turn off loop control, disable STP (where possible) and restart SNMP if you are using it.

Give those a shot and see if there is any improvement. One other thing, remember to try and keep all your switches up to date, especially if you have some with older firmware. It is possible that they could be introducing issues into the network. It's best to keep all your switches at a consistent version if possible.

[dshef]

STP off, seems to be a flat line on memory usage now, we will keep rolling out suggested changes and I will follow up if issues persist!

Thank you so much,
David

[Cole.Linsley]

We have been ahving the same issue with this Memory Leak all switches are on either 1.5.11 and 1.5.14 but we were wondering whats in common between our switches that are rebooting and it seems like its Board Version B. I have seen switches on other Board Versions we have that are not rebooting even if Memory Usage is at 100%. We have been temporarily disabling and reenabling SNMP on the switches which clears the memory after re-enabling SNMP.

On Board Version B we have two switches that we completely disabled SNMP and Discovery on and we have not seen the issue with it rebooting happen since we did that.

Is anyone else seeing this issue on Board Version B? Just curious for our own research

[Cole.Linsley]

Has anyone else had issues with Board Version B on the WS-26-400-IDC rebooting, It seems to be tied to that Board Version we have switches on other Board Versions so I am thinking it is a issue between the software and hardware on that specific Board.