[jira] [Created] (HADOOP-16184) S3Guard: Handle OOB deletions and creation of a file which has a tombstone marker

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

[jira] [Created] (HADOOP-16184) S3Guard: Handle OOB deletions and creation of a file which has a tombstone marker

JIRA jira@apache.org
Gabor Bota created HADOOP-16184:
-----------------------------------

             Summary: S3Guard: Handle OOB deletions and creation of a file which has a tombstone marker
                 Key: HADOOP-16184
                 URL: https://issues.apache.org/jira/browse/HADOOP-16184
             Project: Hadoop Common
          Issue Type: Sub-task
          Components: fs/s3
    Affects Versions: 3.1.0
            Reporter: Gabor Bota


When a file is deleted in S3 using S3Guard a tombstone marker will be added for that file in the MetadataStore. If another process creates the file without using S3Guard (as an out of band operation - OOB) the file still not be visible for the client using S3Guard because of the deletion tombstone.

The whole of S3Guard is potentially brittle to
 * OOB deletions: we skip it in HADOOP-15999, so no worse, but because the S3AInputStream retries on FNFE, so as to "debounce" cached 404s, it's potentially going to retry forever.
 * OOB creation of a file which has a deletion tombstone marker.

The things this issue will cover:
 * Write a test to simulate that deletion problem, to see what happens. We ought to have the S3AInputStream retry briefly on that initial GET failing, but only on that initial one. (after setting "fs.s3a.retry.limit" to something low & the interval down to 10ms or so to fail fast)

 * Sequences
{noformat}
1. create; delete; open; read -> fail after retry
2. create; open, read, delete, read -> fail fast on the second read
{noformat}
The StoreStatistics of the filesystem's IGNORED_ERRORS stat will be increased on the ignored error, so on sequence 1 will have increased, whereas on sequence 2 it will not have. If either of these tests don't quite fail as expected, we can disable the tests and continue, at least now with some tests to simulate a condition we don't have a fix for.

 * For both, we just need to have some model of how long it takes for debouncing to stabilize. Then in this new check, if an FNFE is raised and the check is happening > (modtime+ debounce-delay) then it's a real FNFE.

This issue is created based on [~[hidden email]] remarks and comments on HADOOP-15999.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]