Uptime of 366 days for my linux machine at work

F

fireburster

[H]ard|Gawd
Joined
Mar 29, 2002
Messages
1,265
This is my 40gb mp3 server i use at work thats a amd 2ghz pc i brought in for folding also. It gets daily use thats for sure. Running kanotix and needs updated im sure, atleast a bigger hdd.

366uptime.jpg
 
Neat. What are you doing to push the load to almost 3?
 
streaming mp3s with foobar. Its just sharing mp3s on my network between whatever pc im working on.
 
Ah, right. I use a network share for that. :)
(Though I've got a shoutcast server installed for when I'm not on my local network.)
 
Impressive uptime, best for me was 106 days also on Kanotix, until an extended power outtage drained my backups.
 
Nice uptime. If you let it, it will likely stay up until a motheboard capacitor goes bad.

I remember reading somewhere about a server that had gotten walled off during a building renovation, and wasn't rediscovered until a temporary wall was demolished to make way for some cabling... about five years later. I believe it was running some Novell network software, don't remember which operating system.

My personal best was 419 days for my file server. Runs Gentoo Linux. I shut it off to upgrade the memory and then the kernel.
 
I had an uptime of over 2 years. RH8 originally. Hardly touched it but for software updates. Was mail, www, samba, jabber and a few other things. No local logins allowed, so I got to be selective when it came to kernel upgrades.

What finally brought it down wasn't anything it was doing wrong, specifically. The windows AD at the time decided to stop authing the box. No rhym or reason for it; it just stopped accepting it's auth. Tried everything I could think of, finally reseting the box in case the kernel got corrupted in memory somehow ( wasn't it ). Never did figure out why that windows AD stopped accepting it. Drove me up the wall.
 
I just hit 236 days on one of my systems. It'd probably be close to a year if I hadn't unplugged it to move it across the room at the beginning of the year. It's even survived my friend tripping over it and physically breaking the case without going down. 366 is pretty impressive though.... hopefully mine makes it that far.
 
My machines never do more than 30 days uptime, then again, that is because I take them all(three) to a Lan party every month. I must say, having a file server endure 1TB of OUTBOUND transfer in a 12 hour period once a month makes me proud!

I can't wait for this weekend, 60 hours of leeching, can't wait to see how the system holds up to that punishment. You just can't break this linux box!
 
And how many kernel exploits are out for that system? :D

My SELinux box:
hoka@smirnoff ~ $ uptime
00:14:46 up 321 days, 12:49, 1 user, load average: 0.00, 0.00, 0.00
hoka@smirnoff ~ $ uname -a
Linux smirnoff 2.6.11-hardened-r15 #1 Wed Aug 24 23:17:18 Local time zone must be set--see zic manu i686 Pentium III (Coppermine) GenuineIntel GNU/Linux

Fully open to the world with root remote enabled about 80% of the time, and 0 hacks. Runs apache/rsync/squid.

:D
 
I had my home server at 332 days once.

"Sun Jul 4 15:00:37 CEST 2004
15:00:40 up 332 days, 12 min, 1 user, load average: 0.27, 0.20, 0.14"

Saved that as a memento, then took it down for upgrades (IIRC)
 
Best uptime I've had on my machine has to be around 250+ days. Had to take it down to move.

Linux helium 2.6.14-hardened-r5 #1 SMP Sun Mar 12 08:10:56 GMT 2006 i686 Pentium III (Coppermine) GenuineIntel GNU/Linux
 
hokatichenci said:
And how many kernel exploits are out for that system? :D

My SELinux box:
hoka@smirnoff ~ $ uptime
00:14:46 up 321 days, 12:49, 1 user, load average: 0.00, 0.00, 0.00
hoka@smirnoff ~ $ uname -a
Linux smirnoff 2.6.11-hardened-r15 #1 Wed Aug 24 23:17:18 Local time zone must be set--see zic manu i686 Pentium III (Coppermine) GenuineIntel GNU/Linux

Fully open to the world with root remote enabled about 80% of the time, and 0 hacks. Runs apache/rsync/squid.

:D
Click to expand...
and with a 15 minute load average of 0, this machine just sits there?

I wouldn't be too proud to tell anybody that I had root remotely accessible. period!
 
draconius said:
I wouldn't be too proud to tell anybody that I had root remotely accessible. period!
Click to expand...

For a while the Hardened Gentoo (sub) project had a box open to the world running SELinux to show off how secure it is.

The root password was advertised on the webpage. The box was so tightly secured root couldn't even run /bin/ls.
 
uname -a

linux mastnix 2.6.11-kanotix-7 #1

almost took it down today to put xbuntu on it.
 
doh said:
For a while the Hardened Gentoo (sub) project had a box open to the world running SELinux to show off how secure it is.

The root password was advertised on the webpage. The box was so tightly secured root couldn't even run /bin/ls.
Click to expand...

Correct, on my SELinux system root is setup as "just another user". There isn't anything they can do to mess with the system. You need role sysadm_r to do anything, which logging in directly to root can't get.

Sample id logged in from hoka/newroled to sysadm_r/su to root:
uid=0(root) gid=0(root) groups=0(root),1(bin),2(daemon),3(sys),4(adm),5(tty),6(disk),7(lp),8(mem),9(kmem),10(wheel),11(floppy),20(dialout),26(tape),27(video) context=hoka:sysadm_r:sysadm_t

Sample id of directly logged in root:
smirnoff ~ # id
uid=0(root) gid=0(root) groups=0(root),1(bin),2(daemon),3(sys),4(adm),5(tty),6(disk),7(lp),8(mem),9(kmem),10(wheel),11(floppy),20(dialout),26(tape),27(video) context=root:user_r:user_t
smirnoff ~ # su
-bash: su: command not found

SELinux is crazy for security. As for the load averages, the services it is running arn't exactly intensive, but even if one was exploited they are so locked down that they wouldn't be able to damage the system. The whole idea for the project is to have a highly protected set of local caching so that people can't inject some fake packages into our portage rsync and whatnot. So far its worked great.
 
fireburster said:
streaming mp3s with foobar. Its just sharing mp3s on my network between whatever pc im working on.
Click to expand...

So is "Foobar" an application loaded on the OS to share out and stream the MP3's or something else?
:confused:

I too would like to share out some music files via a Debian box...
 
We run two Debian boxes at a small business I do consulting work for. They were both up for 342 days before I did some hardware upgrades.
 
Carloswill said:
So is "Foobar" an application loaded on the OS to share out and stream the MP3's or something else?
:confused:

I too would like to share out some music files via a Debian box...
Click to expand...

I'd suggest putting samba on it and setting up a windows share. On the client(s), just map the share to a drive letter, check the box for automatic remounting, and use files as if they were local.

Foobar is a music player, AFAIK?
 
foobar is a windows music app yes. I just have the music on a share on the linux box and i stream to my windows xp work machine.
 
I have a box running novell network for a hardware shop and has an uptime of about 5 years thanks to the no-breaks :D

oldmx
 
I'd have an uptime longer than 30 days if my server didn't generate so much heat :p
Another reason for me to start my silent and cool server pc.
 
hells yeah for long uptimes on linux/unix boxes.

I generally fiddle with my servers too much to really rack up any extensive uptimes...I think the longest I've had was about 60 days on my ventrilo / ftp / apache / occasional CS:S server last year.
 
Carloswill said:
This message has been deleted by odoe. Reason: Flamer
Click to expand...

I thought his comment was funny, like the gentoo-is-rice site.
 
I generally get nervous if my machines are up continuously for more than ninety days or so. I start looking for security announcements I may have missed, etc. I think of boxes with huge uptimes as a liability, not something to be proud of.
 
[H]EMI_426 said:
I generally get nervous if my machines are up continuously for more than ninety days or so. I start looking for security announcements I may have missed, etc. I think of boxes with huge uptimes as a liability, not something to be proud of.
Click to expand...
yeah, I completely agree with you
 
[H]EMI_426 said:
I generally get nervous if my machines are up continuously for more than ninety days or so. I start looking for security announcements I may have missed, etc. I think of boxes with huge uptimes as a liability, not something to be proud of.
Click to expand...

I'm kinda with Hemi on this one. Production servers I run for the most part are never up more then 2 months. They are rebooted for updates and whatnot. Also a reboot can be good for it. Personal best on systems I have setup but didn't then have to administer was like 400 and some days on a linux system, 1.5 years or so on bsd, 2 years on a true64 box. The scarry part is my record was a windows 98 lite box I setup as a network mp3 player. Sucker ran almost 3 years without a reboot. It cut off in the end because of a bad powersupply. I've also had my home bsd server running like 300 days. Power outage ended its run.
 
[H]EMI_426 said:
I generally get nervous if my machines are up continuously for more than ninety days or so. I start looking for security announcements I may have missed, etc. I think of boxes with huge uptimes as a liability, not something to be proud of.
Click to expand...

why must you reboot to update the box to fix security issues? This isn't windows we're dealing with...the only thing you ever really need to reboot linux for is a kernel change. On my servers that are up all the time I generally set up cron jobs to update the world packages on a regular basis.
 
Eva_Unit_0 said:
why must you reboot to update the box to fix security issues? This isn't windows we're dealing with...the only thing you ever really need to reboot linux for is a kernel change. On my servers that are up all the time I generally set up cron jobs to update the world packages on a regular basis.
Click to expand...
My rules for updating FreeBSD are as follows...

If I can patch and rebuild applications in the base system, do so without updating the world. If the thing requires a kernel patch and reboot, do so. However, if I get more than four or five patch versions out of date for the OS (remember, FreeBSD is a complete OS, not just a kernel), I rebuild the world at the next patchlevel release as a matter of principle. For example, I had been patching and rebuilding some things in the base system since FreeBSD 6.1-p2. The box's uname still reported 6.1-p2, even though I had patched pretty much everything that had patches available in the base system. However, the updates yesterday or so were rolled in to 6.1-p6, and I hadn't done a mass rebuild since -p2, so in case I missed something I went ahead and did a rebuild of the base system (world) plus the kernel after updating the world sources and rebooted.

The vast majority of the time I don't reboot the machine without a good reason...Like I said before, most of my updates are just built from ports or patches made available by the FreeBSD security team to take care of things in the base system...However, sometimes you have to patch the kernel and reboot.

There are no "world packages" on FreeBSD...Unless you replace everything in the base system with their ports equivalents, but that's a management task all its own.

I never once said that I had to reboot to fix a security update. All I said was that if I go too long without a reboot I start to think I missed something. Don't put words in my mouth. :)
 
[H]EMI_426 said:
My rules for updating FreeBSD are as follows...

If I can patch and rebuild applications in the base system, do so without updating the world. If the thing requires a kernel patch and reboot, do so. However, if I get more than four or five patch versions out of date for the OS (remember, FreeBSD is a complete OS, not just a kernel), I rebuild the world at the next patchlevel release as a matter of principle. For example, I had been patching and rebuilding some things in the base system since FreeBSD 6.1-p2. The box's uname still reported 6.1-p2, even though I had patched pretty much everything that had patches available in the base system. However, the updates yesterday or so were rolled in to 6.1-p6, and I hadn't done a mass rebuild since -p2, so in case I missed something I went ahead and did a rebuild of the base system (world) plus the kernel after updating the world sources and rebooted.

The vast majority of the time I don't reboot the machine without a good reason...Like I said before, most of my updates are just built from ports or patches made available by the FreeBSD security team to take care of things in the base system...However, sometimes you have to patch the kernel and reboot.

There are no "world packages" on FreeBSD...Unless you replace everything in the base system with their ports equivalents, but that's a management task all its own.

I never once said that I had to reboot to fix a security update. All I said was that if I go too long without a reboot I start to think I missed something. Don't put words in my mouth. :)
Click to expand...

I wasn't putting words in your mouth, I just honestly wanted to know why you rebooted the systems periodically instead of doing rolling updates on-the-fly.

I have very little experience with FreeBSD so I'm definitely not going to argue against your system...plus, you have far more system admin experience than myself (as I'm just a normal desktop user with a few household servers and such) so I'm sure your methods are far more sophisticated than mine are.

I just have cron jobs that update the system (I run gentoo, so I can, for example, have the box run an emerge --sync and an emerge --update world at regular intervals) though perhaps this shows my ignorance, not my craftiness.
 
There's nothing wrong with that approach, I'm guessing. I just subscribe to a "don't update unless you've got a good reason to" philosophy, which is right in line with my one of my favorite life philosophies: "if it ain't broke, don't fix it." FreeBSD has three types of branches, -CURRENT, the bleeding-edge version, -STABLE, the "we won't break it on purpose" branch, and -RELEASE, the "patches and critical fixes only" branch. I track -RELEASE on machines I care about. Therefore, things like the system compiler, etc. don't get version bumps until I move to a newer major or minor version -RELEASE. Updates of -RELEASE are just patchlevels, like 6.1-RELEASE-p6, which indicates that this is the sixth patch release of the 6.1-RELEASE branch. I occasionally track -STABLE on some machines in order to fix the ports I maintain, but that's about it.

Does Gentoo have a "release" branch?
 
um...well, the way gentoo handles packages is considerably different from how FreeBSD does it. By default, there are two options: stable and testing. However, it doesn't specifically have a "patches and critical fixes only" branch, though you could create such a scenario by simply masking package versions higher than the ones you want. You could probably also use some sort of overlay to accomplish that.

You have an incredible amount of control (on a package-by-package basis) regarding what versions of packages you run...but it takes effort to set it up according to your needs if you want something other than "omg this might not compile" or "these packages are very stable."
 
Eva_Unit_0 said:
um...well, the way gentoo handles packages is considerably different from how FreeBSD does it. By default, there are two options: stable and testing. However, it doesn't specifically have a "patches and critical fixes only" branch, though you could create such a scenario by simply masking package versions higher than the ones you want. You could probably also use some sort of overlay to accomplish that.

You have an incredible amount of control (on a package-by-package basis) regarding what versions of packages you run...but it takes effort to set it up according to your needs if you want something other than "omg this might not compile" or "these packages are very stable."
Click to expand...
FYI, portage is based off of the FreeBSD ports system.
 
draconius said:
FYI, portage is based off of the FreeBSD ports system.
Click to expand...

yeah I know...but like I've said, I'm not very familiar with FreeBSD so I honestly am not sure how similar they are to use.
 
Ports and pkgs on FreeBSD are a separate entity from the base system (the "world"). Ports (as long as you keep the tree updated, which you should do before you build/install anything new or update anything already installed) are always "current" and don't really have version tagging. They're third-party software and are only somewhat minimally supported by the FreeBSD folks themselves. That's what the port maintainers are for.

I consider ports separate from the base system as far as updates go.
 
[H]EMI_426 said:
.....................................

Does Gentoo have a "release" branch?
Click to expand...

Not at the moment... Sadly...... However there is a GLSA ticket that is being actively developed that would add a patch only branch. It is still being worked on, and has NOT been approved, but is still active, and hopefully someday it will make its way to the portage tree.

One of the arguements against doing this is that it would effectively break "rolling updates" Meaning that the only way top reliably update an older system is to format, and do a clean install.... Then the question also becomes what packages gets included in this branch? How far back should patches be back ported? Where is all this extra man power going to come from? And who is going to pay for it?

There is still alot of uncertainties about going this route.

Right now they have what is essentialy the equal to CURRENT.... anything that starts with a tilda, for example ~amd64 ~x86 ~ppc and so on..... They refer to it as testing.

They also what is more or less the equal to STABLE, which would be anything without a prefix, for example amd64 x86 ppc and so on.... They call this stable

And they have the unstable branch... This is the branch where they add new packages that either flat out dont work, or need to be tested on a package by package basis.... this is anything that stats with a dash, for example -amd64, -x86 -ppc.... They call this unstable.
 
You must log in or register to reply here.
Back
Top