path length issues with CITGM and modules

[Trott]

Not sure what to do about this, but we should solve it somehow. @nodejs/build @nodejs/citgm-admins @mcollina

CITGM results for fastify fail when a socket path name gets truncated. See fastify/fastify#1856 (comment).

fastify shouldn't have to protect against this in their tests in my opinion.

citgm probably shouldn't have to worry about it either.

Can we shorten the temp path generated in the citgm-smoker job?

[richardlau]

We're currently explicitly setting a temp path of "$WORKSPACE"/citgm_tmp everywhere apart from AIX (where it's /ramdisk0/citgm for performance reasons). I don't know the reason why it has been set to be in the workspace.

[Trott]

We're currently explicitly setting a temp path of "$WORKSPACE"/citgm_tmp everywhere apart from AIX (where it's /ramdisk0/citgm for performance reasons). I don't know the reason why it has been set to be in the workspace.

Perhaps to avoid it colliding with temp paths used in other jobs?

[rvagg]

I don't know the reason why it has been set to be in the workspace.

I'd guess because we don't have any guarantees about there existing another unique directory on all of the hosts we may want to use? Maybe we just replace that with ${HOME}/node-tmp like for NODE_TEST_DIR? or a subdir of that dir. It's on all of the machines.

[richardlau]

I'll switch the job over to using a directory in NODE_TEST_DIR later today (when I can keep an eye on a test CITGM build with the change).

[richardlau]

Made the change: https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/jobConfigHistory/showDiffFiles?timestamp1=2019-10-02_19-32-13&timestamp2=2019-10-16_18-40-33

Test build: https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/2059/

[richardlau]

Looks promising:

https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/2056/nodes=ubuntu1604-64/testReport/(root)/citgm/fastify_v2_10_0/history/

[richardlau]

No fastify failures or any new failures that look to be related to the new temp path. I believe this is resolved.

[Trott]

No fastify failures or any new failures that look to be related to the new temp path. I believe this is resolved.

I agree. That's fantastic! Thanks!