cancel
Showing results for 
Search instead for 
Did you mean: 

Any recomendation about reboot Aix Server?

Former Member
0 Kudos

Hi all,

I have many LPAR´s with more than 600 days uptime. Any problem to continue not rebooting the virtual or physical Power Servers? No problems about performance or other cause, just know about best practice.

Thanks,

Carlos Alfredo

Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos

Hey Carlos!

I don't see a problem with NOT rebooting.  We run over 200 LPARs at our company v770s all on AIX 6.1.  over 45 of those are totally dedicated to SAP application instances / oracle DBs.  We only reboot when we absolutely need to.  I would not worry about it!

--NICK

Former Member
0 Kudos

Thanks Nick.

Carlos Alfredo

Answers (2)

Answers (2)

Matt_Fraser
Active Contributor
0 Kudos

I remember once going to an "SAP on Microsoft" conference at Microsoft's Redmond campus. Somebody actually let the data center folks who run Microsoft's own SAP systems give a presentation, and the question about rebooting came up. The speaker said "well, we know there are gradual memory leaks, so we proactively reboot our systems monthly to avoid problems." That was quite an admission!

Granted, though, that was in 1997 and they were running SAP on Windows NT 4.0 and SQL Server 6.5 (as was I at the time). Things have changed and matured considerably since then. My Windows servers can go many months between reboots these days, and as Siddhesh mentioned, pretty much the only reason for a reboot today is if I've applied a security patch that requires it. If it weren't for those patches, they would happily go much longer.

AIX is a different beast, of course. Back in 1997 I also ran an AIX server (though I am no expert) -- in fact it was the system that we were replacing with SAP. We didn't reboot it very often, partly because it took so long to come back up, and partly because none of us fully understood at the time what to do if something went wrong.

Former Member
0 Kudos

Unix servers in general can be up for many hundreds of days and for many administrators, uptime is a badge of honor.  Many servers have gone years with without rebooting and updating servers. Unfortunately, I am from the camp that thinks that not patching Unix servers because they are "Safe" is misguided. In today's security environment, I highly recommended keeping server patch levels up to date and rebooting them when necessary. Yes, even if you ruin your uptime statistic.

With that said, if you can find out what process is getting stuck or causing issues, killing it and restarting the process usually works. 

former_member185954
Active Contributor
0 Kudos

Hello Matt,

so true when you said the following:

We didn't reboot it very often, partly because it took so long to come back up, and partly because none of us fully understood at the time what to do if something went wrong.

I have worked on AIX for major part of my career and yes I agree that its pretty stable, but then IBM keeps sending these TL patches where we are required to constantly have them updated.

IBM stopped support for many older versions of OS lately, so you won't get patches for them but as I mentioned earlier, its all down to what the OS admins implement as a company wide strategy.

SAP kernel is pretty stable and doesn't show any signs of weariness or memory leaks etc unless documented in SAP notes but that too would be version specific ones, it just goes on and on

Regards,

Siddhesh

Matt_Fraser
Active Contributor
0 Kudos

Totally agree. There's no badge of honor here, it's about what is best for the system and for the business. The business doesn't like downtime, but all systems require periodic maintenance, so it's a tradeoff. The business likes being hacked even less, and it's a mistake to think that something isn't vulnerable because it isn't Windows. Unix in fact is just as vulnerable, and Siddhesh's example of the OpenSSL Heartbleed bug affected many more Unix systems than Windows systems.

Former Member
0 Kudos

Justo to mention that are many Unix OS patchs that are applied on the fly, we dont need to reboot de server.

I think that main bugs are in the sap/oracle stack, again we dont need to reboot the server to keep than uptodate.

Carlos Alfredo

former_member185954
Active Contributor
0 Kudos

Hello Carlos,

Would that mean you haven't applied any OS patches that require a reboot . Although there is an old saying 'If it isn't broken, don't touch it' , I wonder how much it holds true in today's interconnected world where lots of changes occur in connected systems/frameworks/landscapes that would push us to take remedial action.

I would suggest that you speak to OS Admins, who should receive regular periodic updates from OS vendors with respect to patches. These patches keep the servers up to date, especially security patches (if any) should be applied regularly.

For eg. There have been recent reports about issues with OpenSSL having a security vulnerability(heartbeat bug etc), which required an urgent patch and potential downtime.

Regards,

Siddhesh

Former Member
0 Kudos

Hi Siddhesh,

Thanks a lot, We know that security concerns are legitime, and I think Unix box are, in general, less exposed to vulnerabilities.

Carlos Alfredo