***Do you perform downtimes just for the purpose of rebooting the systems? ***Is there a recommended interval Linux system should be rebooted? -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.centos.org/pipermail/centos/attachments/20050602/189d0dbf/attachment-0004.html>
> -----Original Message----- > From: centos-bounces at centos.org [mailto:centos-bounces at centos.org] On > Behalf Of Prasad Pillarisetti > Sent: Thursday, June 02, 2005 11:42 AM > To: centos at centos.org > Subject: [CentOS] Reboots > > > ***Do you perform downtimes just for the purpose of rebooting thesystems? It's a matter of policy. We do not but all of our customer facing systems are load balanced. If they weren't and the reboot wasn't of an emergency nature then yes, we would reboot during a maintenance window.> ***Is there a recommended interval Linux system should be rebooted?Ideally never. You should only need to reboot for hardware or kernel changes. We have internal machines that have been up for over 3 years without interruption (let the uptime contest begin). -- Marc
On Thu, 2005-06-02 at 22:11 +0530, Prasad Pillarisetti wrote:> ***Do you perform downtimes just for the purpose of rebooting the > systems? > ***Is there a recommended interval Linux system should be rebooted?Personally I find that workstations need more downtime than servers, but that "downtime" usually consists of logging out and back in again. Anyways, servers running Linux can easily have an uptime measured in years. Only hardware issues tend to bring them down. -- Ignacio Vazquez-Abrams <ivazquez at ivazquez.net> http://centos.ivazquez.net/ gpg --keyserver hkp://subkeys.pgp.net --recv-key 38028b72 -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: This is a digitally signed message part URL: <http://lists.centos.org/pipermail/centos/attachments/20050602/ba060f71/attachment-0004.sig>
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Thu, Jun 02, 2005 at 10:11:30PM +0530, Prasad Pillarisetti wrote:> ***Do you perform downtimes just for the purpose of rebooting the systems? > ***Is there a recommended interval Linux system should be rebooted?Yes. I perform regular reboots on my Linux systems, with scheduled downtimes, each year. All kidding aside, the main reason for it is to use a compressed air machine on all computers, to keep their interior clean. []s - -- Rodrigo Barbosa <rodrigob at suespammers.org> "Quid quid Latine dictum sit, altum viditur" "Be excellent to each other ..." - Bill & Ted (Wyld Stallyns) -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (GNU/Linux) iD8DBQFCnzmvpdyWzQ5b5ckRAhc4AJ4lp+F98JeHPaBW6t5jAgj/sdNQ6ACffEV3 EVeegNxIED1x19tSBkUYIZI=4QLN -----END PGP SIGNATURE-----
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Prasad Pillarisetti wrote: | ***Do you perform downtimes just for the purpose of rebooting the systems? | ***Is there a recommended interval Linux system should be rebooted? Prasad, You shouldn't need to reboot your linux box under normal circumstances. I personally only reboot my linux server when every other option to fix an issue has failed, or I am booting to a new kernel. 11:56:03 up 47 days, 3:12, 1 user, load average: 0.13, 0.04, 0.01 The above is my centos server's uptime. Unfortunately, there was an electrical outage at my home, and the server went down. =-( - -- Alex White prata at kuei-jin.org Fingerprint = 58DC 9199 CE73 74E8 B2C1 442E ACF5 92E0 E068 C46C gpg key location: http://www.kuei-jin.org/GPG-KEY-PRATA ~From the withered tree, a flower blooms --Zen Proverb -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.6 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org iD8DBQFCnzrbrPWS4OBoxGwRArM8AKCIjXBHCJcsLVrjVm3Ijs6QuVIjUwCfU1ez 6tNQ2Hh7LPwoIOOAZH3pels=vypk -----END PGP SIGNATURE-----
Prasad Pillarisetti wrote:>***Do you perform downtimes just for the purpose of rebooting the systems? >***Is there a recommended interval Linux system should be rebooted? > >In my experience, a Linux system must be rebooted in the following circumstances: 1. The kernel (especially one with a security patch) is upgraded 2. glibc is upgraded and the individual instances of old versions in RAM are impacting performance (mostly from kernel processes that cannot be restarted unless one does a reboot) 3. Non-hot swappable hardware failure 4. A kernel module (usually a proprietary hardware driver) is malfunctioning in some way I'm sure I can come up with more circumstances, but the jist is: 1. If the hardware is fine, you will not *need* to reboot. 2. If the kernel is updated, you must reboot to use the new kernel. 3. If glibc is updated, you must reboot to have many processes use the new glibc. Hope this helps, --Shawn
From: Rodrigo Barbosa <rodrigob at suespammers.org>> That, my friend, is very much true. > Depending on the kind of server you are running (mumber of different > processes), a reboot will be much easier. Did that myself a few times.As far as examining running processes, one should have _good_ operational procedures and practices to track what is and isn't running on your system. If you can, stick with Sys-V start/kill init scripts for everything, and document anything that can't be (as well as why). I don't like to rely on reboots to do such things. In fact, when building a new, standard system configuration, I like to shunt to "init 1" and back to "init 3" (run-levels assume distro is Fedora-based) several times to ensure that all my start/kill init scripts are proper. That way I can stop and start user-space services piecemeal and completely. -- Bryan J. Smith mailto:b.j.smith at ieee.org
> ***Do you perform downtimes just for the purpose of rebooting the systems? > ***Is there a recommended interval Linux system should be rebooted? >Personal preference here, besides all the things mentioned on other posts. The "POST" routine can give you early warnings of pending hardware failures that might go un-noticed until complete failure otherwise. Scheduled maintence reboots if for no other reason should be done. my 2?
From: Rodrigo Barbosa <rodrigob at suespammers.org>> Also, run-levels are standarized, and should be the same on all unix-like > platforms.??? Debian and Solaris run-levels differ from Red Hat. And some distros, like Solaris and SuSe, also have a "rcS.d" directory that runs in addition to, and before, the run-level. -- Bryan J. Smith mailto:b.j.smith at ieee.org
From: Ken Godee <ken at perfect-image.com>> The "POST" routine can give you early warnings of pending > hardware failures that might go un-noticed until complete > failure otherwise.In such cases where that is a worry, I'm going to want to deploy real-time hardware management instead. It's worth the extra few bucks. -- Bryan J. Smith mailto:b.j.smith at ieee.org