| [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] |
Hello, Thank you very much to everybody who responded (hall of fame below). The general consensus is that the only time reboots are required is after hardware or kernel upgrades -- at which point they're unavoidable anyway. However, rebooting every once in a while, just for the sake of "peace of mind" is not uncommon. One notes (paraphrased): The nice thing about *nix systems compared to Windows is that you typically have the choice: You can reboot if you want but you usually don't have to. (As an aside, I updated a W2K server this past weekend, using Windows Update. It took six required reboots before I was done.) Some people find that after upgrading shared libraries (e.g. libc), a reboot is a healthy practice. Others have kept systems going for months after library upgrades without any issues. Almost everybody is satisfied with upgrades using the RPM system; they almost always work out well. Several stated that after installing a new package, it's good practice to bring it up "by hand" the first time (e.g. "service whatever start"), to see if the new software behave well -- if it doesn't then it probably won't at boot-time either, and by that time you'll probably have forgotten everything about the context, or will be under a good deal of pressure to bring the box up again quickly. One notes that the reboot strategy depends on the electrical environment: Is the power supply to the box stable (perhaps secured with a UPS) or not? If it isn't, then "unexpected" reboots will happen; those reboots will almost always be un-monitored, so the downtime can be significant if something has gone wrong. (My personal view: If the box is important, put a UPS on it. No matter how much your electricity supplier boasts about perfect performance and what have you. I sleep better that way.) Finally, one suggests that it may be healthy to perform a backup before rebooting after an upgrade; he lost a machine when an update went wrong. Some memorable quotes: * "Personal opinion: you're still locked into the M$ mindset." * "But screw pride. Reboot when you have the maintenance window." *** Hall of fame *** Angel L. Mateo Anthony E. Greene Brad Kittredge Brett Geer Brian Dunbar Bruce Richardson Darrell Parlee Dave McFerren Evan Cofsky Gordon Dey Jim Westbrook Joshua Daniel Franklin Kevin Morland Lance Levsen Mark Price Martin Schmitt Michael McCafferty Neil Dombrowski Raffael Marty Red Lloyd Sean Lamb The anonymous system administrator from Sweden Thank you all! /Martin. On Tue, 2002-04-09 at 13:55, Martin Eskildsen wrote: > Hello, > > The setting is this: A RedHat 7.1 Linux development/file server for one > of our teams has an uptime of around 6 months. During that time, various > RPMs have been upgraded due to various reasons -- security, stability, > bug fixes, you name it. But the box hasn't been rebooted. > > In a way, I'm kinda "proud" and happy about that. I love stable systems > with long uptimes. > > However, I'm also getting a bit worried: What if one of those upgrades > means that rebooting the box will be a pain? It's not impossible to > imagine that one or other package messed something up that won't show > until reboot-time. > > OK, I'll bite the bullet this coming weekend and bring the server down > and up again, just to make sure. > > But I was wondering what you guys do regarding updates that do not > explicitly require rebooting the server: Do you reboot nonetheless, or > do you keep your fingers crossed and hope everything's all right (like > I've done)? > > Thanks, > > > /Martin. > > -- > Martin Eskildsen > Tpack A/S > Mail: mae@tpack.net > Tel.: (+45) 44575011 > WWW : www.tpack.net _______________________________________________ LinuxManagers mailing list - http://www.linuxmanagers.org submissions: LinuxManagers@linuxmanagers.org subscribe/unsubscribe: http://www.linuxmanagers.org/mailman/listinfo/linuxmanagers
[Home] [Kernel List] [Linux SCSI] [Video 4 Linux] [Linux Admin] [Yosemite News] [Motherboards]