Hi all, is there a difference between physicly unplugging a switch and using the reload command ? By difference I mean maybe some important pre-reload process that occur when I issue the reload command.
I ask this question because I have to upgrade many many 3550 (critical production switch) from 12.1.13 to 12.1.22. There's a scheduled blackout in our building so my strategie is to erase current ios version from flash, download ios 12.1.22 on all switch flash couple days before the blackout, configure boot system and wait the blackout. When power will return, all of my 3550 will run 12.1.22.
Should I espect some problems ?
We also have problems getting regular maintenance windows for IOS upgrades and use a similar strategy, which hasn't failed us (so far!).
I see no problems with using the blackout to your advantage as long as you test out and document your upgrade procedure on a test switch first.
This will hopefully ensure that you don't forget an important step or upload a corrupt IOS image to all of your switches.
There is no fundimental difference, both will run the POST etc.
Sometimes an upgrade of this nature removes of renames commands, which automaticaly get deleled from the config. I cannot remember the precise details but I had an issue wity an upgrade of two 3550 switches running enhanced image and HSRP. One of the HSRP priority or policy commands was removed which meant a few minutes head scrating, the IOS did report the change, but only through the console, not in the log.
Other than that the only real issue is that if there is a problem with one ( or more ) of the switches then it may be blamed on the blackout by people who are not aware of the change.
When I have had to upgrade a number of devices like this, I have always tested a single switch before reloading the lot. The timed reload helps a lot with this, i.e. reload at 3:00 am on a sunday morning when nobody in thier right mind is at work. Just be carefull that you don't interupt a network backup cycle.
So there's no config recompilation, memory cleanup or whatever other process/script lauched before the reload command reboot the switch.
We are not upgrading to the latest version for stability reasons, those switch are very critical and we are short on time and we don't have time to do intensive test. Staying in the same train seems safer for me.
No, the config file is read from NVRAM at boot time by the new IOS operating system. If there are any config modifications necessary due to deprecated or removed commands in the config they get changed then.
If you want to be 100% sure, load the config on a "spare" switch in a lab envionment and upgrade and then reboot. Assuming of course you have the necssary hardware available.
I would often do this, I had a campus with about 300 switches and poor power.
I would do a load from ciscoworks, reboot a couple, make sure I had no issues, then load the rest... Wait for a month or so, then check to see which ones hadn't reloaded.. then ask for the change request to do those select few.
12.1.13 is pretty ancient version of ios for the 3550. Any reason your not updating to the 12.2.35 code?
I upgraded a couple of 3550 and 3500XL using ciscoworks. I did test one each switch model to run at 3 AM and it worked fine. Again RPAQUIN is right if you have a change Mngt in place, I would open up a change request. One you upgrade all the switches, you can actually schedule a reload or a blackout :-) You can also upgrade and then reload afterwards. If you have no ciscoworks, I would do it manualy on each switch and wait for the black out to happen all the config will stay the same.
actually, if he does have ciscoworks, I'd be very leery with 12.1.13... I remember something about Cisco changing the directory structure/how the update worked from 12.1.13 to 12.1.19.. something that Ciscoworks wasn't very happy about. That was a few years ago, my memory could be shot, but I'd definitely test that. If I remember right, after doing a xmodem upload on several, I decided a manual update to 12.1.19 then updating from there was safer.
No worries if your doing the putty, paste and 18 thousand windows approach. Just make sure you use the solarwinds tftp server or something multithreaded..not the old cisco one.
My labs teste are conclusive so I'll go this way : manually update switch with SecureCRT-3CDaemon, wait blackout and check for hardware / software failure when power is back.
I have just finished upgrading 1,060 3550's. I would greatly stress that you do not delete the old image file if at all possible. You run the risk of a corrupted IOS file or something of that nature and it takes forever to upload a new IOS via a console connection. If you leave the old file on there all you have to do to recover is console in and change the boot statement, then reload.