In the last poll we discovered some 90% of responders run MySQL on Linux which is not totally a surprise. Number of people asked me to do the followup pool to see what specific Linux distributions are most popular in MySQL space hence this is the poll for today.
Please answer what Linux Distribution you’re using for your production environment. You can select more than one. If your Linux distribution is missing please ether select the closest in the list or add the comment so I can add it to the pool.
What Linux Distributions do you use for running MySQL
- CentOS (33%, 418 Votes)
- Ubuntu (24%, 302 Votes)
- Debian (22%, 277 Votes)
- RHEL (9%, 115 Votes)
- Gentoo (4%, 44 Votes)
- Fedora (2%, 31 Votes)
- Oracle Linux (2%, 21 Votes)
- SLES (2%, 19 Votes)
- Slackware (1%, 15 Votes)
- OpenSuse (1%, 8 Votes)
- Mint (0%, 3 Votes)
Total Voters: 971
Loading ...
I included CentOS, but we’re migrating away from it to Debian. 6.x has been laughably unstable on VMWare for some reason.
Why no Slackware?
mainly CentOS 6.x but starting to like Oracle Linux 6.x with UEK kernels
We run it on CentOS, but going to migrate to Ubuntu.
Jacket,
OK I added Slackware though I have not come across a lot of people running MySQL in production on it this is why I have not included it initially.
Thanks! I voted. I’m just curious, because Slackware is really widely used in my country. I’m using it on all of my servers.
I run it on CentOS. Since Oracle decided to be nice with their distro, I’ve wanted to test it out with Galera. I hate trying to run anything on Debian based distros, just because of what I find to be annoying.
Once you added “Slackware”, the WP-Polls plugin reset the maximum number of allowed votes to “1” – you will want to fix this.
An annoying bug.
Shlomi,
Thanks fixed. Annoying bug indeed.
hi guys,
So, to everyone considering moving from CentOS – would love to hear some real tangiable reasons, eg: ‘unstable on vmware’ is just FUD unless backed up with some bug reports….
And, if there is anything that we, the CentOS side of things, could be doing better to make it a more enticing mysql host platform, do tell..
K Singh
CentOS Project
Karanbir
For me, some of the reasons at looking at other OSes (Oracle Linux 6.x) would be more updated packages. Kernel and another off top of my head would be openssl 1.0.1c (SPDY/Google Elliptic Curve Diffie-Hellman support would be nice). iuscommunity.org repo is working on updated package https://bugs.launchpad.net/ius/+bug/1034961
Yeah not exactly mysql specific, but for me at least servers aren’t just for mysql servers but entire web server stack.
I voted for RHEL and CentOS, but would add another RH derivative – Amazon Linux. If you’re in AWS anyway, it’s relevant, and it’s definitely not stock RHEL or CentOS.
Now, if only we could get Percona Server in RDS… is there any chance of that partnership happening?
K Singh,
My apologies; I’m certainly not trying to FUD it up around here. The nature of what we’ve been seeing has made it almost impossible for me to report a bug on this particular issue. I’m a firm believer in only making bug reports that have reasonable diagnostic and don’t sound like magic. But, here’s the nature of what I’ve been *seeing* for the last year, and why I ran out of options. We moved from RH to Cent in dev/staging, it was clean in staging for a few months, so I pushed it to prod. Then the restarts began.
6.0 (obviously out of date, but moving MySQL servers in a ‘zero downtime’ environment takes a *long* time to happen).. basically what happens is that lightly loaded boxes, disproportionately running MySQL and MongoDB, just restart. Sometimes it’s a “nice” restart with a memory dump (which fails because these box have more RAM than disk space in /var). Other times it just disappears and comes back. It vary rarely happens on RW masters. I did a quick count late this week. We have ~40 MySQL tiers in production (m-s-s triples, so ~120 total). Of those, I have one sharded setup with mostly identical VMs that I use as my production variance testing. There are ten shard masters, running various distributions and MySQL versions (5.1 and 5.5, commercial, GPL, and Percona). Cent 6.0 and 6.2 boxes reboot about every 2-3 months.
So, while I wish I could give you solid bug reports, I simply don’t have any. But for the sake of my sanity, I had to trust my limited statistics and do the Cent -> Debian migration across the board…after almost completing the RH -> Cent migration in it’s totality.
in fairness, though, when I reported the SNMP/XFS regression is 6.0, it was promptly fixed. I don’t have anything against CentOS the distribution, and certainly not the people behind it. But I can’t just take product outages, and I don’t have time to track it down correctly.
Best,
Wes
We run hundreds of different customers and DB servers almost exclusively on CentOS 5 and now a bit of 6, with most running on top of Xen 3, and solid hardware (but also AWS, Chinese clouds, hardware). We almost never have OS or serious DB problems (we get about one crash per year), fortunately (well, there is a Percona/Innodb DB crash this week on altering a partitioned table, but that’s another thing).
We have nothing but problems with Debian/Ubuntu and try to avoid them when possible, but we are very RH/Centos/RPM-focused for our own sanity, and system stability.
Happy to share our base build lists, how / where we get solid latest Centos/RH packages for everything in common stacks or tools (always a real challenge) and default configs.
I use Ubuntu and CentOS, but I prefer Ubuntu.
I saw few comments from people saying they are migrating from CentOS, not sure why?
We been running CentOS to process millions of phone calls, and we have number huge databases and we never had an issue with OS.
@Dean, I like CentOS, but from a sys admin standpoint when using the command line, I am more confident with Ubuntu than I am with CentOS. With CentOS, I always fear that I will screw something up.