Community discussions

MUM Europe 2020
 
jo2jo
Forum Veteran
Forum Veteran
Topic Author
Posts: 967
Joined: Fri May 26, 2006 1:25 am

remote reboot = dangerous?

Fri Oct 06, 2006 8:55 pm

do any of you guys do reboots on remote MT devices? i'm starting to think i took this for granted. but now out of only 10-13 remote reboots over teh past year, i've had 3 times when the device never came back up and required a power cycle...two times this happened on 2.9.24 and now one on 2.9.30

Luckly the MT stuff is SO solid i NEVER have to reboot (i have APs with 100s of days of uptime and never any issues..these all serve laptop wifi ppl), but when i do, do a reboot i expect the device to come back, or is this common knowlage that you dont do remote reboots unless you MUST and i missed the memo?

i did check mac-ping, telnet ect..too

tks
joe
 
GotNet
Member
Member
Posts: 436
Joined: Fri May 28, 2004 7:52 pm
Location: Florida

Fri Oct 06, 2006 9:17 pm

We do it a lot. But mostly for upgrades. In four years, it has only failed once.

Mike
 
User avatar
jp
Long time Member
Long time Member
Posts: 600
Joined: Wed Mar 02, 2005 5:06 am
Location: Maine
Contact:

Fri Oct 06, 2006 9:24 pm

don't have problems with RBs doing remote reboot. I did once, but that was because someone left a serial cable plugged into the console.

PCs once in a while will have a setup issue or bad bios/battery and not reboot properly. No fault of routerOS. "Keyboard error - press F1 to continue."
 
jo2jo
Forum Veteran
Forum Veteran
Topic Author
Posts: 967
Joined: Fri May 26, 2006 1:25 am

Sat Oct 07, 2006 10:26 am

well luckly about 1 1/2 min before i did this reboot which never came back, i did a supout for a different issue, so i emailed it to support to see if they can see what happened...not that i think its anything specific, but this rb has done this before.

just for their own records really..
 
User avatar
Equis
Forum Veteran
Forum Veteran
Posts: 888
Joined: Mon Jun 06, 2005 6:48 am

Sat Oct 07, 2006 2:58 pm

I find routeros 99.9% come back up.

I put the watchdog on to be sure if something goes wrong but this will not always help.

1 time out of 100's it never came up and I had to reflash it (stood in front with my coat open)

Apart from that all very good. :shock:
 
jo2jo
Forum Veteran
Forum Veteran
Topic Author
Posts: 967
Joined: Fri May 26, 2006 1:25 am

Sat Oct 07, 2006 6:41 pm

yea i had/have watchdog on...well luckly (or not :) ) the location has iffy power and looses it about once a month now for the past year...last time was sept 18...so lets keep our fingers crossed :) i have UPS on everything EXCEPT this device so that good....everything else runs off PoE switches so i can always do very easy remote reboots..

with my luck ill be waiting months for a power outage..

tks

edit: oh and i have yet to have any flash related issues either.
 
jarosoup
Long time Member
Long time Member
Posts: 600
Joined: Sun Aug 22, 2004 9:02 am

Sat Oct 07, 2006 7:37 pm

We've had this happen on wrap.2c boards. Upon a power cycle, they can get a errors on disk, requiring the disk repair - this is when they fail to reboot. Some flash cards do this more than others (Lexar, PQI, PNY have all lost our trust). Never seen this problem on Soekris or Routerboard.
 
User avatar
sten
Forum Veteran
Forum Veteran
Posts: 920
Joined: Tue Jun 01, 2004 12:10 pm

Sun Oct 08, 2006 2:17 am

We've had this happen on wrap.2c boards. Upon a power cycle, they can get a errors on disk, requiring the disk repair - this is when they fail to reboot. Some flash cards do this more than others (Lexar, PQI, PNY have all lost our trust). Never seen this problem on Soekris or Routerboard.
that's because routeros doesnt load the cpu specific kernel modules, to support the cpu specific reset instructions, before issuing the reboot. to make wrap work in this manner you need to add additional software interrupts to the wrap bios (or tamper with the loading sequences, but i *do not* recommend that) to get it to work right.
the disk error can be something as simple as buffer on the CF was not completely written to do disk before the power was gone (this is not a big issue for the integrity but must be handled before mouting a journaling filesystem to not overwrite the pending journal entries). almost all CF cards default to DMA mode transfers and wrap 2c does not support DMA which can easily trigger relativly harmless errors like that.
Move along. Nothing to see here.

Who is online

Users browsing this forum: gibi13, Znevna and 151 guests