WS-12-250-DC units constantly quickly rebooting

[HSC]

Hi all,
I could not find an e-mail address to contact support... posting diagnostic info in public forums is risky due to possibility of revealing internal network details, so I will limit the info that I include here.

We saw an issue with a site that has a WS-12-250-DC switch (EF:9E), this issue kept occurring for about 20 minutes after the Netonix was switched out for another unit of the same model (12:90) that had the same config, which was restored from a backup file from the first. Both were running 1.5.25. The issue stopped while I was troubleshooting it onsite. I wish to know the reason to prevent it from happening again.

Issue: constant rebooting approx 1-2, sometimes every 10 minutes after the switch had previously restarted powering on. Happened for about 2 hours until I got onsite.
I was trying to login to the web interface but it would only show a very slow "Loading..." bar and usually the switch would reboot by the time it hit ~65%.
I connected via SSH and was reviewing the log but did not see anything interesting up to just before the reboot.

Note: this switch occasionally (every 2-3 months on rough average, has been online since a year) saw full-site reboots before, on the rare occasions that it would fall to battery after loss of rectifier/charger power (meaning drop from ~54V to ~50V). It has huge newer ~100AH batteries so definitely would not see notable voltage drop from the Netonix's 2-3A load moving on the batteries. In case this fall onto batteries was triggering the "hibernate" function, I changed that from the industry-average 41V to the battery-damaging-risk 30V a few months ago, but it's uncertain if that fixed the issue of the reboots when falling into batteries. During the above issues, the site was on batteries for few minutes but then back on the charging 54V.


Other threads that I checked, and reasons that they do not apply:

Periodic reboot of WS-12-250-DC switches viewtopic.php?f=17&t=8207
-not related to our issue:user had ancient firmware and 60 days between reboots
-our SNMP server(s) may poll more than once per minute, I have not checked recently, but that is minor CPU usage on anything not from the 1990s.
-switch is hooked up directly to 48V battery bank with circuit breaker

WS-26-500-DC REBOOT ISSUE EXPLAINED viewtopic.php?f=6&t=3621
-internal cabling issue causing I2C errors, not related to our model

Netonix Boot Loop Issue When Powering ePMP 3000L Radios viewtopic.php?f=17&t=8252
-our ePMP3000 APs are using 48V PoE

WS-12-250-ac reboots every 5 minutes viewtopic.php?f=17&t=8242
-thread suspects an OS issue but no specific conclusion

WS-8-150 self reboot viewtopic.php?f=17&t=8154
switch randomly rebooting viewtopic.php?f=17&t=8087
-seems to be an old firmware issue

Spontaneous Simultaneous Reboots viewtopic.php?f=17&t=7391
-(R)STP / possibly LAG related, which we don't need here/have disabled

WS-12-250-AC rebooting like crazy viewtopic.php?f=17&t=6974
-unrelated, seems some AC PSUs had bad caps

Thank you.

[sirhc]

We do not offer support except on the forums, if we need something sensitive we will provide you an email for contact.

Periodic reboot of WS-12-250-DC switches viewtopic.php?f=17&t=8207
-not related to our issue:user had ancient firmware and 60 days between reboots
-our SNMP server(s) may poll more than once per minute, I have not checked recently, but that is minor CPU usage on anything not from the 1990s.
-switch is hooked up directly to 48V battery bank with circuit breaker

Actually the CPU is only there to run the UI, LACP, RSTP, SNMP, NTP, SSH. All packet switching is handled in the switch core which can handle line rate on all ports at the same time. So a large CPU is not needed. In fact the CPU in this SOC is the very same mips processor that was in the Linksys WRT545G which was released in December 2002. Again we did not design the SOC and pick the CPU much smarter people who work for MicroChip of of the largest chip maker in the world.

Providing the full MAC address is NOT a security issue so why so cryptic. To even attempt to talk to the switch via MAC you would have to be in the local flat network segment and not sure our switch would respond to direct hardware address communications.

Note: this switch occasionally (every 2-3 months on rough average, has been online since a year) saw full-site reboots before, on the rare occasions that it would fall to battery after loss of rectifier/charger power (meaning drop from ~54V to ~50V). It has huge newer ~100AH batteries so definitely would not see notable voltage drop from the Netonix's 2-3A load moving on the batteries. In case this fall onto batteries was triggering the "hibernate" function, I changed that from the industry-average 41V to the battery-damaging-risk 30V a few months ago, but it's uncertain if that fixed the issue of the reboots when falling into batteries. During the above issues, the site was on batteries for few minutes but then back on the charging 54V.

Do you have the switch hooked DIRECTLY to the battery bank or to the charger unit?

Are you using SFP modules?

Post up screen pics of ALL Tabs that have changes. You may block out sensitive info like Lat/Long, IP address, SMTP Host and email address, and no on.
NOTE: FULL MAC IS NOT A SECUIRTY ISSSUE AND I NEED IT.

Upload screen grabs do not point to 3rd party host.

upload.png (3.79 KiB) Viewed 150 times