LinuxQuestions.org
Register a domain and help support LQ
Go Back   LinuxQuestions.org > Forums > Linux Forums > Linux - Software
User Name
Password
Linux - Software This forum is for Software issues.
Having a problem installing a new program? Want to know which application is best for the job? Post your question in this forum.

Notices

Reply
 
Search this Thread
Old 01-02-2012, 04:31 PM   #1
Jim4522
LQ Newbie
 
Registered: Apr 2010
Location: Salisbury Vermont
Posts: 17

Rep: Reputation: 0
Does anyone know how many times the average instance of Linux is rebooted a year?


I am not sure if this is a software question but does anyone know how often the average server operating under Linux needs to be rebotted per year or per month?

Is it likely that a server machine would operate for a year and never be rebooting for any reason?

Jim White
 
Old 01-02-2012, 04:37 PM   #2
T3RM1NVT0R
Senior Member
 
Registered: Dec 2010
Location: Internet
Distribution: Linux Mint, Ubuntu, SLES, CentOS
Posts: 1,790

Rep: Reputation: 324Reputation: 324Reputation: 324Reputation: 324
@ Reply

Hi Jim4522,

Interesting question!!!

I am not sure if anyone else will be able to give a straigh answer to this. As far as I am concerned I would say that there is no such figure that I have. I mean I am not sure how many times a server will require a reboot. Usually a reboot is require in the following situations:

1. When server is running low on memory.
2. When server encounters any fault.

The above will depend on the number of applications running on the server. Utilization of that particular server etc. So there is no fixed number or there is no average number available.

A server can run for a year or more or may require a reboot every day/week depends upon its usage.
 
Old 01-02-2012, 04:39 PM   #3
TobiSGD
Moderator
 
Registered: Dec 2009
Location: Hanover, Germany
Distribution: Main: Gentoo Others: What fits the task
Posts: 15,592
Blog Entries: 2

Rep: Reputation: 4046Reputation: 4046Reputation: 4046Reputation: 4046Reputation: 4046Reputation: 4046Reputation: 4046Reputation: 4046Reputation: 4046Reputation: 4046Reputation: 4046
The only reasons I can think of for a server to reboot are either hardware maintaining or a kernel update. Besides that I don't know why to reboot. I even do it rarely on my home systems, most of the time I use suspend or hibernate.
 
Old 01-02-2012, 04:45 PM   #4
gilead
Senior Member
 
Registered: Dec 2005
Location: Brisbane, Australia
Distribution: Slackware64 14.0
Posts: 4,123

Rep: Reputation: 151Reputation: 151
The operating system itself is robust enough to just keep running. I've had boxes running over 400 days and there are plenty of sites that count uptime in years.

It depends what you're running on top of the OS. Does it leak memory or use hard disk space over a period of time? I don't know whether you you can get averaged stats across Linux boxes generally, but google brings back plenty of people who post their info
 
Old 01-02-2012, 06:10 PM   #5
Nominal Animal
Senior Member
 
Registered: Dec 2010
Location: Finland
Distribution: Xubuntu, CentOS, LFS
Posts: 1,723
Blog Entries: 3

Rep: Reputation: 942Reputation: 942Reputation: 942Reputation: 942Reputation: 942Reputation: 942Reputation: 942Reputation: 942
I fully agree with T3RM1NVT0R, TobiSGD, and gilead, above.

The servers I use are only rebooted after upgrading the kernel due to security patches. Even those could be avoided using Ksplice or perhaps kexec, but rebooting is simpler. Those are rare, say once or twice a year on average, some years not at all.

However, I have a number of workstations I use regularly, but not every day. On this one, the SMART data on my hard disks indicate about 700 power cycles in 11620 hours of use, or about 16 hours per power cycle on average. That matches my estimates: the motherboard I used until recently did not suspend well, so I tended to shut down the computer practically every night. (Actually, I also have a physical switch I easily flip to disconnect my workstation, including monitor and all peripherals, from mains power. The previous display I had wasted a lot of watts in standby, so it was very handy. Normally the SMART power cycle count will also include the count of long periods of inactivity, but in my case, those are extremely rare, certainly less than 10% of the power cycle count.)

Because of this, I don't think there is any sensible way to define an average instance of Linux. It would be like saying that an average human speaks Mandarin Chinese. It is true, technically, but it is not very useful in determining typical human characteristics. To define typical Linux characteristics, for example reboot rate, you'll need to first guess, estimate, or analyse the different typical Linux instances, and to group them in interesting groups, before any averaging... Perhaps separating embedded/utilities, servers, and workstations/desktops?

Last edited by Nominal Animal; 01-02-2012 at 07:58 PM. Reason: added .. 'a year' to the second paragraph
 
Old 01-02-2012, 06:29 PM   #6
fukawi1
Member
 
Registered: Apr 2009
Location: Melbourne
Distribution: Fedora & CentOS
Posts: 854

Rep: Reputation: 189Reputation: 189
Personally, my home gateway machine, a HP DL380g2, has only been powered down as a result of local power failures.
If it wasn't for the blackouts, it would have been up for 2.5-3yrs non stop. I have yet to have a situation where I had to reboot it.
I even blow the cobwebs out of it with a compressor and vacuum every few months, while its still up.

Although, I have to disagree here..
Quote:
1. When server is running low on memory.
There is a bunch of stuff you can do to fix this before rebooting.
Quote:
2. When server encounters any fault.
and here, I'm guessing you mean a hardware fault, but in case of software faults, again, there are many things you can do before resorting to a reboot.

I would say that requiring regular/frequent reboots of a Linux server, would be indicative of something being wrong with it.
 
Old 01-02-2012, 08:09 PM   #7
jlinkels
Senior Member
 
Registered: Oct 2003
Location: Bonaire
Distribution: Debian Lenny/Squeeze/Wheezy/Sid
Posts: 4,087

Rep: Reputation: 492Reputation: 492Reputation: 492Reputation: 492Reputation: 492
I have some servers running at 700+ days. I am scared to death to reboot them, you never know what hardware failure developed over time.

jlinkels
 
Old 01-02-2012, 09:29 PM   #8
frankbell
Guru
 
Registered: Jan 2006
Location: Virginia, USA
Distribution: Slackware, Mageia, Mint
Posts: 7,722

Rep: Reputation: 1459Reputation: 1459Reputation: 1459Reputation: 1459Reputation: 1459Reputation: 1459Reputation: 1459Reputation: 1459Reputation: 1459Reputation: 1459
One of the members of my LUG found a server at his workplace that had been up for over three years (I say and he said "found" because it had been running trouble-free for so long that the staff had forgotten about it).

I have one computer at home (the "downstairs" computer) which I commonly shut down every day, primarily so it's not drawing electricity over night. The others get rebooted when I decide to reboot, probably on an average of once a month. The one I'm on right now has been up for slightly over two days, because it just got a new kernel.
 
Old 01-03-2012, 08:03 AM   #9
Jim4522
LQ Newbie
 
Registered: Apr 2010
Location: Salisbury Vermont
Posts: 17

Original Poster
Rep: Reputation: 0
Thank you all for your responses. I think knowing how many reboots servers experience is an important statistic for those who manage IT organizations of all sizes. It is a measure of the stability of servers that can be compared over time and from one IT organization to another and I consider that comparison provides insight into how well IT organizations are managed. Three years ago I did an audit of a major global bank that operated approximately 10,000 servers in just one major city in the US. This was prior to the current recession and that bank had hired one maintenance vendor to provide 24/7 on-site resident support for all problems arising from those 10,000 servers at a cost exceeding $2,500,000 per year. That vendor provided monthly reports on every problem it responded to every month. Over half of the “problems” were requests from “users” to the help desk to reboot “their” server for reasons that made sense to the “user”. Of the other problems reported to the help desk the vender’s first response for almost half of them, out of laziness I suspect, was to reboot the server to see if that solved the problem before doing anything else. In some cases that resolved the problem and in others it didn’t. For the problems that were re-reported to the help desk the vendor performed a fix and then reboot the server/machine. Of those fixes less than 50% actually solved the problem which meant the vendor had to try a different fix at a different time which also included a reboot. In all I counted over 48,000 reboots on those 10,000 servers in just one year and that didn’t include the reboots that were automatically performed by failover systems or other automated processes, thus my question. Jim White
 
Old 01-03-2012, 01:43 PM   #10
T3RM1NVT0R
Senior Member
 
Registered: Dec 2010
Location: Internet
Distribution: Linux Mint, Ubuntu, SLES, CentOS
Posts: 1,790

Rep: Reputation: 324Reputation: 324Reputation: 324Reputation: 324
@ Reply

@ fukawi1,

When I said server running low on memory I was pointing to memory fragmentation. I was describing about the situation where your server ran out memory and is unable to allocate memory to other applications. No doubt you can try different things but memory fragmentation is normal and it happens with machine whether it is linux or windows it doesn't matter. I do agree that it happens more with Windows machines but linux is no exception. Memory fragmentation totally depends on the applications running on the server.

When I said server encounters a fault I was pointing to server coring not necessarily an hardware issue.

Quote:
I would say that requiring regular/frequent reboots of a Linux server, would be indicative of something being wrong with it.
It is not necessary that if a linux server require a regular or frequent reboot there is something wrong with it. What if I have a linux server configured as primary ldap server and everyday 5000 users and other devices on the LAN perform LDAP queries to this server everyday. It is quite obvious that one day this server will go down on the resources and will require a reboot to regain it.

Another example will be of an http servers. What if you have a single http server for your parent and child domain and everyone is perform http requests throughout day. Obviously it will require a reboot after few days.

The thing that I want to point here is there is nothing wrong with any machine. The main thing is how you design your network and how you load balance your services and how you use these machines.

We cannot answer the question how many times a server require a reboot unless we know how the network is designed, services that are running on a server, number of users it will going to serve etc.
 
Old 01-03-2012, 09:35 PM   #11
jefro
Guru
 
Registered: Mar 2008
Posts: 11,527

Rep: Reputation: 1404Reputation: 1404Reputation: 1404Reputation: 1404Reputation: 1404Reputation: 1404Reputation: 1404Reputation: 1404Reputation: 1404Reputation: 1404
We have lots of servers that could run a year or two but random things usually happen like fire in room or ups service or updates. You never seem to keep them up as long as they could run.
 
Old 01-03-2012, 09:59 PM   #12
Jim4522
LQ Newbie
 
Registered: Apr 2010
Location: Salisbury Vermont
Posts: 17

Original Poster
Rep: Reputation: 0
Giving all of the above is it more or less likely that the average server operated by an IT organizations will experience at least one reboot a month or one a year?

Jim White
 
Old 01-03-2012, 10:13 PM   #13
gilead
Senior Member
 
Registered: Dec 2005
Location: Brisbane, Australia
Distribution: Slackware64 14.0
Posts: 4,123

Rep: Reputation: 151Reputation: 151
In organisatiions I've worked in, production servers are more likely to be up for several months at a time. Even in those places, I've seen administrators who reboot servers once a month, just in case.

I don't think you'll be able to successfully generalise across multiple organisations. Part of the reason is that uptime isn't the goal, the implied stability is - and that can be achieved in several ways. Load balancing, failover, etc provide continuity of service.
 
Old 01-03-2012, 11:08 PM   #14
Nominal Animal
Senior Member
 
Registered: Dec 2010
Location: Finland
Distribution: Xubuntu, CentOS, LFS
Posts: 1,723
Blog Entries: 3

Rep: Reputation: 942Reputation: 942Reputation: 942Reputation: 942Reputation: 942Reputation: 942Reputation: 942Reputation: 942
Quote:
Originally Posted by Jim4522 View Post
It is a measure of the stability of servers that can be compared over time and from one IT organization to another
No, it is not. Linux reboot rate has nothing to do with stability. (Except for closed source drivers, and drivers still in testing stage; those do tend to crash the system now and then.)

Linux machines are rebooted because the admin/user wants to, not because they need to.

Quote:
Originally Posted by Jim4522 View Post
and I consider that comparison provides insight into how well IT organizations are managed.
Perhaps. With respect to Linux servers, I basically agree.

However, "effectively never" means to me the IT organization is not at all interested in security matters. "Every month" indicates the organization most likely has a Windows background, and is basically working from a crib sheet. The former is usually easy to fix (via gentle education, and scary scenarios). The latter is not.

To me, "whenever there is a kernel update important enough to warrant a reboot", is the best answer. (And then we discuss who and how the security issues are monitored.)

Quote:
Originally Posted by Jim4522 View Post
Giving all of the above is it more or less likely that the average server operated by an IT organizations will experience at least one reboot a month or one a year?
I'd say once a year sounds likely, to upgrade the kernel (and/or storage drivers, if using vendor-provided drivers). This has nothing to do with stability, and everything to do with security.

(If you use vendor-provided drivers not in the mainline Linux distribution, then it is advisable to upgrade whenever the vendor recommends. I guess that is kind of a stability issue. I simply prefer to use hardware with vanilla kernel.org Linux kernel support instead.)

I've excluded all external causes for reboots from my consideration -- floods, fire hazards, cooling problems in server halls, lack of electricity, lack of connectivity -- since those need to be estimated on a case-by-case basis anyway.
 
Old 01-04-2012, 06:51 AM   #15
Jim4522
LQ Newbie
 
Registered: Apr 2010
Location: Salisbury Vermont
Posts: 17

Original Poster
Rep: Reputation: 0
Two additional questions for this tread:

Do you think virtualization increased or decrased the "Reboot Rate" of distributed server machines? (the "Reboot Rate" is the average number of machine days between reboots for a given IT organization, the higher the rate the less frequently reboots are be initiated)?

Do you think Windows or linux has the higer "Reboot Rate"?

Jim White
 
  


Reply

Tags
linux, rebooting, servers


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Inquiry:How to set the Linux server to be auto rebooted at pre-specified times? hadimotamedi Linux - Newbie 4 11-12-2009 06:25 PM
To find, how many times my box has rebooted gsiva Linux - General 5 05-30-2009 02:12 AM
So I just rebooted into Linux.... UMG:Chicken_Sop Linux - General 1 11-12-2007 06:44 PM
Mu linux can not be rebooted cjt0610 Linux - Hardware 7 12-27-2006 12:37 AM
LXer: EE Times Announces Recipients of the Educator of the Year and Student of the Year ACE Awards LXer Syndicated Linux News 0 03-29-2006 02:21 AM


All times are GMT -5. The time now is 03:35 PM.

Main Menu
My LQ
Write for LQ
LinuxQuestions.org is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
Syndicate
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
identi.ca: @linuxquestions
Facebook: linuxquestions Google+: linuxquestions
Open Source Consulting | Domain Registration