Bug #7255
closed[Data Manager] Test some non-trivial manifests in integration test
0%
Description
Currently, the test data used in integration tests include only one data block per manifest, and the test program relies on this: it doesn't test the safety of anything after the first block in the first stream of each collection.
The testing scheme should expand to include the following cases:- Empty collection (
manifest_text == ""
) - Collection with 100 streams with 10 blocks each (total 1000 different blocks)
- Handle errors in manifest format (see #7253)
Updated by Brett Smith over 9 years ago
- Target version set to Arvados Future Sprints
Updated by Brett Smith about 9 years ago
- Assigned To set to Radhika Chippada
- Target version changed from Arvados Future Sprints to 2015-12-02 sprint
Updated by Radhika Chippada about 9 years ago
- Story points changed from 0.5 to 1.0
Updated by Radhika Chippada about 9 years ago
- Status changed from New to In Progress
Updated by Tom Clegg about 9 years ago
I don't think we can rely on arv-put in createMultiStreamBlockCollection. It will only write one block per stream unless we write more than 64 MiB of data; we'd have to write ~60 GiB in order to comply with the story spec. We're only interested in the number of blocks and streams, not bytes and files.
How about something like
var locs []string for s := 0; s < 100; s++ { manifest += fmt.Sprintf("./stream%d ", s) for b := 0; b < 10; b++ { loc, _, err := kc.PutB([]byte(fmt.Sprintf("s %d b %d", s, b))) locs = append(locs, strings.Split(loc, "+A")[0]) manifest += loc + " " } manifest += "0:1:dummyfile.txt\n" } coll := make(map[string]string) arvadosclient.Create("collections", nil, map[string]string{"manifest_text":manifest}, &coll) return coll["uuid"], locs
The difference betwen TestPutAndGetCollectionsWithMultipleStreamsAndBlocks and TestPutAndGetCollectionsWithMultipleBlocks seems to be that one has multiple streams and the other has just one; both have multiple files, but only a single block per stream. There are also some unnecessary complications that seem to negate the value of the test: if the first block appearing in the manifest is the only one that's backdated, for all we know only the rest of the blocks' age, not their presence in the manifest, is keeping them alive.
I think we only need one test here:- Create one collection with 100 streams x 10 blocks each = 1000 different block IDs (but the total data should be small)
- Ensure list of locs returned by collection-creator func has 1000 different entries (sort and then test all adjacent pairs?)
- Write one additional "stray" data block (unreferenced by any collection)
- Backdate all of the blocks
- Run garbage collection
- Ensure the "stray" block is missing (i.e., we didn't just skip GC entirely for some reason), and all others ("locs") are still present.
Updated by Tom Clegg about 9 years ago
LGTM @ e1c0499
really minor nits:- // instead of /* */ for short comments
- TestPutAndGetCollections[...] is a bit odd as a test name. Maybe TestManifestWithMultipleStreamsAndBlocks. Not a big deal.
Updated by Radhika Chippada about 9 years ago
- Status changed from In Progress to Resolved
- % Done changed from 0 to 100
Applied in changeset arvados|commit:efacbb4c548a4d8c9bda4a8b99a1f2f13e15cd2f.