1038014 - panda move train 3, releng portion

Status: RESOLVED FIXED

(Infrastructure & Operations Graveyard :: CIDuty, task)

Comment 1

[hwine]

Attachment: train_3.csv

impacted hosts, fqdn

Comment 2

[hwine]

Attachment: train_3_short.txt

affected hosts, short name only

Comment 3

[hwine]

note: no hosts in slavealloc change names, so no database update scripts generated

Comment 4

[hwine]

all affected host tracking bugs renamed

Comment 5

[hwine]

Attachment: train3-devices.patch

scripted output for train 3 devices.json output

Comment 6

[hwine]

Attachment: last-train-cleanup.patch

Comment 7

[hwine]

Comment on attachment 8455534 [details] [diff] [review]
last-train-cleanup.patch

These are manual find/change on remaining instances of scl1 in the file. Unsure why these were not in the spreadsheet, but there's no more equipment in scl1.

Comment 8

[hwine]

Comment on attachment 8455534 [details] [diff] [review]
last-train-cleanup.patch

https://hg.mozilla.org/build/tools/rev/30445d3c1013

Comment 9

[hwine]

Comment on attachment 8455533 [details] [diff] [review]
train3-devices.patch

https://hg.mozilla.org/build/tools/rev/48875a247c66

Comment 10

[hwine]

per panda directories recreated on foopies via manange_foopies.py

pandas re-enabled in slavealloc

We're done here!

Comment 11

[Pete Moore [:pmoore][:pete]]

We had a tree outage (bug 1039170) and disabled all train 3 pandas. It turned out, we hadn't updated the tools checkout on the foopies to pick up the new devices.json file. So as a note for any future data center migrations of pandas - we should update tools on the foopies after updating devices.json.

Thanks,
Pete

Comment 12

[hwine]

This was a clean oops -- we did that on train 2 (bug 1035167 comment 8), and it's now documented in the wiki: https://wiki.mozilla.org/ReleaseEngineering/How_To/Android_Tegras#To_enable_pandas