Labs Project Tested: planet
Site/Location:<EQIAD|CODFW> i actually don't care, eqiad or codfw is fine with me
Number of systems: 1
Service: planet (planet-venus)
Networking Requirements: internal IP
Processor Requirements: 1
Memory: not much , 1GB ?:)
Disks: 20GB
Other Requirements: misc-web can talk to it , n/a/
Description
Description
Details
Details
Related Changes in Gerrit:
| Subject | Repo | Branch | Lines +/- | |
|---|---|---|---|---|
| Introduce planet1001.eqiad.wmnet | operations/dns | master | +3 -1 |
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | • Cmjohnson | T105510 reclaim zirconium (wipe disk) | |||
| Resolved | Dzahn | T104946 move all misc services from zirconium to a VM | |||
| Resolved | Dzahn | T101730 move planet from zirconium to a ganeti VM | |||
| Resolved | akosiaris | T101899 EQIAD: 1 VM request for planet |
Event Timeline
Comment Actions
Change 217230 had a related patch set uploaded (by Alexandros Kosiaris):
Introduce planet1001.eqiad.wmnet
Comment Actions
IP/DNS assignment for VM planet1001.eqiad.wmnet has been done and the VM created on the ganeti01.svc.eqiad.wmnet cluster. MAC address is: aa:00:00:19:24:b9
The rest of the steps in https://wikitech.wikimedia.org/wiki/Ganeti#Create_a_VM from "Update DHCP" and onwards can be followed now to have the VM installed with an OS. Tentatively resolving this (we 've only run this process once up to now), ping me if you need help so I can fix errors.
Comment Actions
current status was:
https://gerrit.wikimedia.org/r/#/c/220310/2 got merged
I can confirm the instance is there on ganeti1003
Creation time: 2015-06-10 08:09:15
Modification time: 2015-06-23 21:36:56 State: configured to be up, actual state is up
etc, but i could not get a console, just no output
and i can't SSH to it, it did not get an IP yet:
NICs:
- nic/0:
MAC: aa:00:00:19:24:b9
IP: Nonewill debug more on carbon
Comment Actions
getting IP now, installer starts, getting console.
problem then was after installer is finished and instance reboots it goes into a cycle and PXE boots again and again
changed boot order to disk with:
sudo gnt-instance modify -H boot_order="disk" planet1001.eqiad.wmnet
GRUB installation failed │
│ The 'grub-pc' package failed to install into /target/.
Jun 24 00:28:44 in-target: Package 'grub-pc' has no installation candidate
Jun 24 00:28:44 in-target:
Jun 24 00:28:44 debconf: --> PROGRESS STOP
Comment Actions
That error confuses me. It means the for some reason the repo did not have grub-pc (not jub the .deb missing, but the Packages files did not mention grub-pc either), which is weird. Trying to reproduce it I modified boot_order back to network and rebooted the VM. The installation went fine in the very first try. Did one more and then set boot_order back to pxe. The VM is now installed fine and waits for the next steps. I am hesitant to call this a heisenbug yet, but I haven't been able to reproduce.
Comment Actions
@akosiaris thank you very much. afraid i don't have much to add why you could not reproduce it, but happy that it works now