fix(ami): bump packer build volume from 16GB to 20GB (WIP, doesn't fix root cause) - #2339
Closed
brainrake wants to merge 3 commits into
Closed
fix(ami): bump packer build volume from 16GB to 20GB (WIP, doesn't fix root cause)#2339brainrake wants to merge 3 commits into
brainrake wants to merge 3 commits into
Conversation
test-ami-nix (15, amd64/arm64) has been intermittently failing with "No space left on device" during StepProvision, reproduced on unrelated PRs (#2329, #2332) while PG17/orioledb-17 build fine on the same 16GB build volume - the PG15 build closure sits close enough to that limit to tip over depending on small variance in build/cache state.
…urrogate/build volumes The prior 16->20GB bump targeted /dev/xvdc (the scratch build-vol) and left the actual constraint untouched: the amazon-ebssurrogate builder instance boots from its own root disk (source Ubuntu AMI's default /dev/sda1), which is where ansible/apt provisioning actually runs before the chroot into the surrogate root at /mnt. Confirmed via CI logs - "No space left on device" was raised inside ansible's own Python logging flush while writing /tmp/ansible.log, unrelated to xvdf/xvdh/xvdc sizing. Adds an explicit launch_block_device_mappings entry for /dev/sda1.
brainrake
changed the title
On this HVM AMI, /dev/sda1 aliases to the same underlying root device as ami_root_device's /dev/xvda, so sizing it to 16GB actually grew the *output* AMI's root snapshot to 16GB - breaking test-ami-nix for every Postgres version (including 17/orioledb-17, which were passing before) since the test harness launches instances with an 8GB root, which can no longer satisfy that snapshot: "Volume of size 8GB is smaller than snapshot ... expect size >= 16GB". Back to just the xvdc bump (16->20GB) from the prior commit, which is at least harmless even though it didn't fix the original ENOSPC error. Root cause is still open: ansible's own log flush hits ENOSPC on the builder instance's own root during provisioning, and that disk isn't one of the three currently-configured volumes (xvdf/xvdh/xvdc).
brainrake
changed the title
Collaborator
Author
|
Closing as WIP - the xvdc bump here is harmless but doesn't fix the actual root cause (see PR description for the two failed attempts, including a regression that was reverted). Filed the investigation findings as PSQL-1570 for follow-up. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Status: root cause still open
Need to identify which actual EBS volume/device `/tmp` (on the builder instance, pre-chroot) resolves to, without touching whatever device the final AMI's root snapshot is derived from. Suggest not merging until that's nailed down - happy to hand off details.
Test plan