Closed Bug 518966 Opened 15 years ago Closed 15 years ago

create Linux VM for production-mobile-master in 650castro

Categories

(mozilla.org Graveyard :: Server Operations, task)

Product:
mozilla.org Graveyard
Component:
Server Operations
Platform:
x86
Linux
Type:
task
Priority:
Not set
Severity:
normal

Tracking

(Not tracked)

Status:
RESOLVED FIXED

People

(Reporter: jhford, Assigned: phong)

References

Details

We would like to experiment with using a Linux machine as our buildbot master for our mobile work.  Matthew was saying that we have a couple ESX hosts at 650 that we could use.  Could we have 2 Linux VMs, each with 2 virtual cpus and 2gb of ram if there is capacity?  These VMs need to be connectible from the N810s which are on the Mozilla-Build WiFi network.  The buildbot needs to have long lasting (3-4hours) TCP sessions between the master and slaves, are there any firewalls which will disconnect these sessions?
Assignee: server-ops → phong
phong was going to investigate the ESX deployment, I'll double-check on the firewall timeouts.
Would these be running on the office vlan?
I'd say they should be on the Build/QA vlan (500). This circumvents firewall issues.
whatever vlan the wireless network 'Mozilla-Build' is on would be perfect
I just the 2 ESX server we have in the office and they only have 1 GB of RAM each.  I will see if we have spare RAM to increase those ESX servers.
1GB?  They should have 8GB - they used to be build ESX servers!
So it does have 8 GB of RAM.  What OS would you like?  I will need to add additional vlans to the ESX host to support this.
can i have the centos 5 ref platform please?
Phong,

We'll need to reconfigure the ESX hosts to have trunked interfaces. Let me know
if you'd like to do this today, or mrz can help you next week.
dmoore: inventory doesn't have the ports information.  you can track down which port mv-vmware02.office is patched into and add that to the same vlan as the mobile devices or create a new vlan for it.

in the mean time, i will have to pull down the build ref image.
I have the ports identified for mv-vmware01 and mv-vmware02, but neither server is set up for trunking. I'll need to switch the ports to trunk, which may take the servers offline until we reconfigure their interfaces to recognize it.
Going to make the switch changes over the weekend.
ports are:

mv-vmware01 - mv-core1:1/0/29
mv-vmware02 - mv-core1:1/0/31
secondary ports are:

mv-vmware01 - mv-core1:2/0/6
mv-vmware01 - mv-core1:2/0/7
switchports are now configured for trunking, including vlan 500
Just wondering what the status of these VMs is, we are hoping to do the TP4 mobile roll out next week using this VM as the master.
Been having problems copying the ref image over.

Windows says 171 mins left on the transfer...
windows failed, trying once more.  if that fails, are you okay building this from scratch?
the VM? i guess.  I don't have access to VI though so I either need that or I need someone to do a basic install+enable ssh for me.
phong, VM transferred and registered on mv-vmware01.  Doing the same to 02.
What is status of this VM? 

Sorry, I cant figure out from the last few comments what is going on.... this should be a clone of our linux refimage VM, so confused by the comments about "windows failed"?!?
OS: Mac OS X → Linux
Summary: Local Linux VM for mobile-master → create Linux VM for mobile-master in 650castro
(In reply to comment #22)
> We would like to experiment with using a Linux machine as our buildbot master
> for our mobile work.  Matthew was saying that we have a couple ESX hosts at 650
> that we could use.  Could we have 2 Linux VMs, each with 2 virtual cpus and 2gb
> of ram if there is capacity?  

jhford just reminded me we asked for *2*, one for production, one for staging. Can these be called:

production-mobile-master
staging-mobile-master
Summary: create Linux VM for mobile-master in 650castro → create Linux VM for production-mobile-master in 650castro
I should have this up by tomorrow.  Before I clone these, can you logon to the ref VM and clean it up.

centos-5-ref-tools-vm.mv.mozilla.com  10.250.48.134
Phong, i have set that vm up as needed.  I wasn't able to delete scratchbox on /builds or unmount that drive so I was wondering if you could either format that drive or replace it with a blank image and i can make a new filesystem on it.  I also removed the SSH keys, disabled puppet and turned off Xvfb.
john - you have root on there so you could run mkfs on it too, no?
(In reply to comment #23)
> I should have this up by tomorrow.  Before I clone these, can you logon to the
> ref VM and clean it up.
> 
> centos-5-ref-tools-vm.mv.mozilla.com  10.250.48.134

err... what needs to be cleaned up on the ref image? We dont normally need to change anything on the refimage before cloning - is this a sideeffect of the copy from MPT to MV?
I do have root, but I am unable to umount the filesystem.  It is saying that the device is busy even though i can't find any file in /builds, /scratchbox or the device file itself in the output of lsof.

For using this machine as a buildbot master, I am removing scratchbox and the production ssh keys as there is no need for them on these unit test only machines.
sorry, that last sentence should read "on this buildbot master"
are those nfs mounts that it is trying to mount?  I see it trying to mount to some nfs volumes at startup and takes a long time to time out.
production-mobile-master  10.250.48.135
staging-mobile-master.mv.mozilla.com  10.250.48.136
Status: NEW → RESOLVED
Closed: 15 years ago
Resolution: --- → FIXED
Product: mozilla.org → mozilla.org Graveyard
You need to log in before you can comment on or make changes to this bug.