System Deployment Steps:
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | hashar | T95045 install/deploy labnodepool1001 | |||
| Resolved | • Cmjohnson | T95048 labnodepool1001 setup tasks: labels/ports/racktables | |||
| Resolved | • chasemp | T95303 Remove hashar and dduvall root access on to be installed labnodepool1001 | |||
| Restricted Task | |||||
| Resolved | hashar | T102281 Figure out fine sudo rules for the nodepool service / diskimage-builder | |||
| Resolved | hashar | T96426 Allow gearman and zeromq connections from labnodepool1001 to gallium.wikimedia.org | |||
| Resolved | hashar | T96867 Use systemd for Nodepool |
Change 202446 had a related patch set uploaded (by RobH):
setting the mgmt and production dns entries for labnodepool1001
Change 202446 merged by RobH:
setting the mgmt and production dns entries for labnodepool1001
Change 202450 had a related patch set uploaded (by RobH):
setting labnodepool1001 install params
network switch port setup has an issue, described in sub-task T95048, once resolved installation can continue.
OS is installed, but attempting to sign keys afterwards has lead to an issue. I cannot ssh or ping labnodepool1001.eqiad.wmnet from palladium (puppetmaster). I can do so from carbon.
This seems a bit odd, still investigating.
bastion, carbon, gallium... hosts in public IP vlans can ping the host, but nothing in the private vlans...
chatted with andrew, this is a known thing, and iron can ssh in. resuming installation
I have created a basic Debian package for Nodepool (T89142) and installed it on labnodepool1001.eqiad.wmnet.
For testing purposes I have created a basic configuration file under /etc/nodepool and installed a local mysql server.
On Wikitech I have created a user nodepoolmanager which is an admin of the labs project contintcloud. The nodepool configuration file reflects that user credentials.
The hardware has been allocated for Nodepool, so I removed the blocking task T93706: [SchemaGatherClicks] Missing or empty schema
Most of the puppet patches have been merged, the one left over is the systemd configuration T96867: Use systemd for Nodepool
The service is implemented and managed to magically boot and delete an instance. The labs work made by @Andrew in spring has been a huge benefit. There is still lot of work but I the first phase is now complete: nodepool is up and running.