Riza Suminto has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/17054 )

Change subject: IMPALA-10497: Fix flakiness in 
test_no_fd_caching_on_cached_data.
......................................................................


Patch Set 1:

(1 comment)

Thanks Joe for the feedback.
I will revert most of these and leave only the extra warm up part.

http://gerrit.cloudera.org:8080/#/c/17054/1/tests/custom_cluster/test_hdfs_fd_caching.py
File tests/custom_cluster/test_hdfs_fd_caching.py:

http://gerrit.cloudera.org:8080/#/c/17054/1/tests/custom_cluster/test_hdfs_fd_caching.py@190
PS1, Line 190: cluster_size=1
> I think this should not impact the caching. Our scheduling algorithm maps f
I will remove this so the cluster size back to default.
Previously, when I kept cluster size as default (3), after several iteration of 
looping this test, one impalad failed with the following error.

I0210 17:50:00.781529 10340 mem_tracker.cc:83] Creating tracker 
/tmp-sharded_lru_cache->root
I0210 17:50:00.782321 10340 status.cc:129] Failed to delete old cache file 
/tmp/impala-cache-file-fc4060e3ab7e45b0:a680f0d3b05005a1: Not found: 
/tmp/impala-cache-file-fc4060e3ab7e45b0:a680f0d3b05005a1: No such file or 
directory (error 2)
    @          0x2c008d7  _ZN6impala13GetStackTraceB5cxx11Ev
    @          0x1c9003e  impala::Status::Status()
    @          0x36016f9  
impala::io::DataCache::Partition::DeleteExistingFiles()
    @          0x3601e22  impala::io::DataCache::Partition::Init()
    @          0x360975e  impala::io::DataCache::Init()
    @          0x35b4db0  impala::io::DiskIoMgr::Init()
    @          0x251b229  impala::ExecEnv::Init()
    @          0x29a0412  ImpaladMain()
    @          0x1bdf4d3  main
    @     0x7faa8e3bebf7  __libc_start_main
    @          0x1ae81da  _start
I0210 17:50:00.897989 10340 mem_tracker.cc:87] Destroying tracker 
/tmp-sharded_lru_cache->root

I was guessing that since all three write their data cache to common /tmp/ dir, 
one might accidentally delete cache file that belong to other impalad. Just 
running the test one time doesn't reveal this error.
So in this patchset 1, I follow examples from custom_cluster/test_data_cache.py 
to set custom_cluster=1.



--
To view, visit http://gerrit.cloudera.org:8080/17054
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I774f9dfea7dcc107c3c7f2b76db3aaf4b2dd7952
Gerrit-Change-Number: 17054
Gerrit-PatchSet: 1
Gerrit-Owner: Riza Suminto <riza.sumi...@cloudera.com>
Gerrit-Reviewer: Bikramjeet Vig <bikramjeet....@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <impala-public-jenk...@cloudera.com>
Gerrit-Reviewer: Joe McDonnell <joemcdonn...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <riza.sumi...@cloudera.com>
Gerrit-Comment-Date: Thu, 11 Feb 2021 02:18:19 +0000
Gerrit-HasComments: Yes

Reply via email to