|
I still can't get an iSCSI storage solution working 
|
#
¿
Jul 19, 2014 00:12
|
|
|
|
| # ¿ Apr 28, 2024 13:29 |
|
|
|
|
#
¿
Jul 19, 2014 00:50
|
|
|
is there a manga guide to high availability storage clusters because I am not getting this to work on my own 
|
|
#
¿
Jul 30, 2014 17:24
|
|
|
|
|
#
¿
Jul 30, 2014 18:07
|
|
|
hello YOSPOS, just wanted to give a lil update re: my high availability storage project I like glusterFS, a lot, I think it's the best solution. unfortunately citrix doesn't think the same so xenserver doesn't support it for what I need. the industry is using DRBD as a backing store for most purposes anyway, and it's directly supported by corosync, so i started barking up that tree again short version: gently caress centos longer version: centos seems to work just fine with DRBD. at first. you can make a virtual disk manually and test it out and everything works fine. then you set up corosync and make an iscsi target and it works fine. then ou try shutting down one of the machines and it fails over just fine. then you start the machine back up and it can't access the partition you set up for DRBD anymore. wtf!!! a lot of searching and apparently it's some kind of race condition in the startup scripts, but nobody's offering any tips to fix. so, after trying out a lot of different configuration crap & following guides I basically figured out how to customize the configuration on my own and got it all working on debian. not only does it work right but debian vastly outperforms centos when syncing the data over the network. so yeah. there's a bit of a peculiarity with debian not having a startup script for tgtd, but you can just add it to rc.local and bob's your uncle. holy poo poo it took me a while to figure out. might actually work better in centos 7 but I couldn't find all the requisite parts for drbd/corosync because EPEL/elrepo havne't caught up yet. idk. I try to use centos because it's the "enterprise" thing to do but debian just works??
|
|
#
¿
Jul 31, 2014 18:24
|
|
|
I would like to thank everyone who made suggestions before & I guess this is a little advanced for this thraed!!
|
|
#
¿
Jul 31, 2014 18:31
|
|
|
also want to point out that I tried openfiler. attempting to update openfiler renders it to an unusuable state. wtf
|
|
#
¿
Jul 31, 2014 18:42
|
|
|
I used gentoo in the mid '00s. never again.
|
|
#
¿
Jul 31, 2014 20:29
|
|
|
well now that I have HA storage I maybe can't get my xenserver pool to properly do its HA stuff because the CPUs don't match. hrm
|
|
#
¿
Jul 31, 2014 20:53
|
|
|
good call
|
|
#
¿
Aug 1, 2014 01:08
|
|
|
well I never got xenserver's HA to work right, which I guess is because it's just not designed to work with less than 3 hosts in the pool?? it says it does but it doesn't, when I tried looking at the log file it just endlessly kept trying to find the other host. but I tried HA-Lizard, which I was wary of at first because it requires some command line config (I have to keep things so monkeys can maintain it) but it does what I need to do (restart a VM on another server if the server goes kablammo) and I can recommend it for smaller set ups!!
|
|
#
¿
Aug 7, 2014 17:09
|
|
|
Um just for clarity. I have 2 surplus HP machines running XenServer configured as a pool. Each is running a Debian VM with DRBD/TGT/Pacemaker to create an iSCSI target with failover capability. Agile VMs can use this as their storage and run on either host. HA-Lizard takes care of restarting VMs if a host goes down. So I have got a redundant cluster that isn't nearly as good as a real solution with like 4 hosts and a dedicated storage cluster but it's better than running an old windows 2000 server under a leaky pipe.
|
|
#
¿
Aug 7, 2014 17:18
|
|
|
well poop, my slave server still won't promote itself to master on its own if the master goes down. idk how to fix
|
|
#
¿
Aug 7, 2014 17:33
|
|
|
this place is already doomed
|
|
#
¿
Aug 7, 2014 17:41
|
|
|
when I tried gluster/nfs xenserver would consistently lose access to the share in a simulated failover. I couldn't find anyone else using it who had posted their config so I gave up on it. iSCSI works just fine
|
|
#
¿
Aug 7, 2014 17:55
|
|
|
xenserver doesn't have glusterfs by default and adding it to the dom0 just mounts it as local storage so you can't use it to run an agile VM anyway I'm doing this because our domain controller is just 1 machine so if it goes down someone has to go to the office and rebuild it from a backup or w/e. we ended up with some spare iron that could run xenserver so I've been going through options, ideally I'd wanna get the shared storage on dedicated hardware when the budget allows
|
|
#
¿
Aug 7, 2014 18:13
|
|
|
idk man I've got 2 test rigs and I need like 3 or more things running, also this is part of my side project of evaluating xenserver as a replacement for vsphere so it's a lil compilcated
|
|
#
¿
Aug 7, 2014 18:22
|
|
|
I work at a little tech school with < 20 employees
|
|
#
¿
Aug 7, 2014 18:33
|
|
|
so yes, unless I go commandeering machines from the classrooms and that is problematic i'm lucky these even have hardware virt
|
|
#
¿
Aug 7, 2014 18:35
|
|
|
reminder that my school literally bought the last batch of pentium 4 optiplexes from dell
|
|
#
¿
Aug 7, 2014 18:36
|
|
|
pram ur a little hostile and it's not helpingEgan Yardley posted:10 years ago tho, right? it was before I started here but it may have been as recent as 5 years ago. they needed more computers and thought it would be a good idea to find the same model they already had
|
|
#
¿
Aug 7, 2014 18:49
|
|
|
well it works out I guess because we have 1 ghost image for all the student machines
|
|
#
¿
Aug 7, 2014 18:51
|
|
|
another thing we do is we have some windows 7 VMs on a vsphere host so students can remote into them. because we needed windows 7 for an office admin class but it's a small class and doing vms was cheaper than upgrading all the pentium 4s
|
|
#
¿
Aug 7, 2014 18:59
|
|
|
did summore loving around with xenserver today, all the HA stuff "just works" if you have identical CPUs but it gets dodgy if if you don't. but it looks like the CPU is the only part it's picky about which is nice because one of our problems w/ vsphere is having to find motherboards with raid/ethernet that's vsphere compatible and that's waaaay less of a problem with xenserver
|
|
#
¿
Aug 12, 2014 20:44
|
|
|
can't find out why but if you're using 2 different CPUs in a pool and the master fails, it just won't let the slave promote to master. I can do it from the command line but it won't do it automatically. but really idgaf because once we get this into production it'll be on the hardware it needs and not just my dumbass test rig made up of old workstations
|
|
#
¿
Aug 12, 2014 20:46
|
|
|
my stepdads beer posted:it can use masking so if they're similar enough everything will just work too I tried this from the start but it just wouldn't autopromote a slave if the master failed. Didn't matter which machine was master. I could promote it manually if I ssh'd to the remaining machine though. HA-Lizard threw an error message along the lines of "this server isn't privileged to become master" despite it being the one selected by the pool to autopromote. couldn't find anyone with the same problem, I'm assuming it's just some really obscure edge case with the two computers I was using to test (they're old core2 HP desktop, one has a slightly newer CPU)
|
|
#
¿
Aug 12, 2014 23:24
|
|
|
I rebuilt my xenserver pool today and now everything just works???
|
|
#
¿
Aug 13, 2014 21:14
|
|
|
|
|
#
¿
Aug 13, 2014 21:15
|
|
|
what's the deal with nginx vs. apache? just in a nutshell
|
|
#
¿
Aug 16, 2014 22:26
|
|
|
they're both webservers right???
|
|
#
¿
Aug 16, 2014 22:26
|
|
|
I don't understand any of that
|
|
#
¿
Aug 16, 2014 22:58
|
|
|
so do you have to run over 9000 virtual machines serving the same content to get apache to maybe work as well as nginx for certain types of site?
|
|
#
¿
Aug 16, 2014 23:15
|
|
|
haven't had any problems with xenserver this week
|
|
#
¿
Aug 20, 2014 02:41
|
|
|
ask smythe
|
|
#
¿
Aug 25, 2014 18:11
|
|
|
turns out the reason my laptop has been running like poo poo lately is because the fan died & the cpu is throttling back when it overheats. $5 shipped please get here fast 
|
|
#
¿
Aug 29, 2014 18:24
|
|
|
my laptop fan is in the mail
|
|
#
¿
Sep 2, 2014 15:55
|
|
|
I got the new fan in my laptop. It's quieter than the old fan.
|
|
#
¿
Sep 3, 2014 16:53
|
|
|
The original macbook fans sounded like a jet taking off. this 3rd party replacement is almost inaudible at full blast (well, with office background noise)
|
|
#
¿
Sep 3, 2014 16:56
|
|
|
ok I finally got the fan to spin up to 6000rpm and I can hear it but it's still like half as loud I guess
|
|
#
¿
Sep 3, 2014 19:06
|
|
|
|
| # ¿ Apr 28, 2024 13:29 |
|
|
ehehehehehe
|
|
#
¿
Sep 28, 2014 19:56
|
|