Wrap paragraphs.

Logs and Sentry links are public now.

Revert "Use install -d for /var/log/ceph."

This reverts commit 8f29b3f9f1f294c449bb5215767cad67fdf7cde6.

Reviewed-by: Warren Usui <warren.usui@inktank.com>

schedule_suite.sh: default email to ceph-qa@ceph.com

Signed-off-by: Sage Weil <sage@inktank.com>

Revert "get_scratch_devices(): catch CommandFailedError, log "dev in use""

This reverts commit 416508de9c628c5cae1b3a7f05bb3ec0e3639bdb.

NameError: global name 'CommandFailedError' is not defined

README: add deps

Reported-by: Loic Dachary <loic@dachary.org>
Signed-off-by: Sage Weil <sage@inktank.com>

s/ceph-debug/ceph-debuginfo for rpms

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

get_scratch_devices(): catch CommandFailedError, log "dev in use"

Signed-off-by: Dan Mick <dan.mick@inktank.com>
Reviewed-by: Sage Weil <sage@inktank.com>

Merge pull request #63 from ceph/wip-logger

Never use 'except:' without specifying an Exception.

Never use 'except:' without specifying an Exception.

Merge pull request #62 from ceph/wip-logger

Add hung job description to email

run: do not import at top level

This makes a run with --lock hang when connecting, for some
reason.  E.g.,

 $ teuthology -v a.yaml --lock
 ...
 INFO:teuthology.task.internal:Opening connections...
 DEBUG:teuthology.task.internal:connecting to ubuntu@plana06.front.sepia.ceph.com
 <hangs>

No clue what is going on here, but this fixes it!

Signed-off-by: Sage Weil <sage@inktank.com>

Add hung job description to email

Merge pull request #61 from ceph/wip-logger

Log all the things! (Things like exceptions.)

Workers: only log child's stderr, not stdout

Don't print colon if there's no exception message. Also, remove testing
exception.

Pretty-up write_initial_metadata; s/run/name/

Add exception hook

run: write info.yaml on job start

Include basic metadata bout the job.

Signed-off-by: Sage Weil <sage@inktank.com>

Move logging setup out of main()

test

run: set up archive dir and logger output sooner

Signed-off-by: Sage Weil <sage@inktank.com>

Merge pull request #58 from ceph/queue

Don't assume anything about the base path here.

schedule_suite.sh: allow partial 'suite' names

Mangle the name (/ -> :) in the run name.  Separate base from
relative path so that the job names don't include a full path.

Signed-off-by: Sage Weil <sage@inktank.com>

suite: return [] for build_matrix base case

This gives us an empty list instead of a crash on an empty dir
or bad symlink

Signed-off-by: Sage Weil <sage@inktank.com>

Merge pull request #60 from ceph/sentry

Add run name to first line of email.

Merge pull request #59 from ceph/wip-collections

redo the way we describe test collections

Add run name to first line of email.

suite: build_matrix as a list of filenames, not file content

Signed-off-by: Sage Weil <sage@inktank.com>

suite: allow individual files to be scheduled

Signed-off-by: Sage Weil <sage@inktank.com>

suite: generalize collection expansion

Previous, a collection was a directory like this:

 mycollection/
 mycollection/facet1/
 mycollection/facet1/1a.yaml
 mycollection/facet1/1b.yaml
 mycollection/facet2/
 mycollection/facet2/2a.yaml
 mycollection/facet3/
 mycollection/facet3/3a.yaml
 mycollection/facet3/3b.yaml

and this would expand to

 1a + 2a + 3a
 1a + 2a + 3b
 1b + 2a + 3a
 1b + 2a + 3b

The fixed directory depth and requirement for a subdir even
when there is only 1 item is annoying.  Instead, allow an
arbitrary directory structure, with the following rules:

 - a .yaml file is a taken as-is (duh); other files still
   ignored
 - a directory is normally just a way to organize files.  We
   recursively descend and build a list of what we fine.
 - a directory with a '%' file in it is special:
     - take the product of every item in the dir (much like
       we did before)
 - a directory with a '+' file in it is special:
     - concatenate everything in the dir into one job

Note that this is equivalent to the previous structure if we
do:

 for facet in mycollection/* ; do touch $facet/% ; done

We can clean up slightly be taking any dir with only one yaml
file in it and replacing the dir with the bare .yaml.

Once this is done, we can reorganize directories however we
like.

Signed-off-by: Sage Weil <sage@inktank.com>

suite: clean up imports

Signed-off-by: Sage Weil <sage@inktank.com>

suite: make collection args (optionally) relative to a base dir

Signed-off-by: Sage Weil <sage@inktank.com>

suite: make names a bit more friendly

Signed-off-by: Sage Weil <sage@inktank.com>

suite: add --dry-run

Signed-off-by: Sage Weil <sage@inktank.com>

Don't assume anything about the base path here.

Merge pull request #54 from ceph/sentry

Add logs and Sentry events to suite emails

Don't return inside __init__

Add apology for non-public links

Indent wrapped exceptions.

s3tests: extend for multi-region tests

Added code to the s3tests task to extract
multi-region info so that that data
can be added to the S3TEST_CONF file
used to run S3 tests.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

rgw: persist the region info

The rgw task deletes the region info
from the config structure. The s3tests
task needs this info, so we persist
it by sticking it in the ctx object.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewd-by: Josh Durgin <josh.durgin@inktank.com>

radosgw-agent: store radosgw_agent server/port

Persist the hostname and port number used
by the radosgw-agent http server.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

move multi_region_enabled() into util class

Moving a helper function into a more general
location so that it can be used by other
classes.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

internal: whitelist btrfs lock imbalance bug 6097

Signed-off-by: Sage Weil <sage@inktank.com>

ceph.conf.template: add osd_debug_pg_log_writeout = true

Signed-off-by: Samuel Just <sam.just@inktank.com>

Merge pull request #55 from ceph/wip-undefined

Reviewed-by: Sage Weil <sage.weil@inktank.com>

Fix undefined symbol errors

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>

Time is an integer, in seconds.

Make email formatting way, way nicer.

Move job listing logic to get_jobs()

Merge pull request #53 from ceph/wip-asok-retry

fix admin_socket retry

admin_socket: fix retry-on-error behavior

Signed-off-by: Sage Weil <sage@inktank.com>

Avoid double slashes in sentry event URL

Add tags! Task name and owner to start.

Add sentry events to suite email.

Also leave a list of sentry events in the summary

Add URL to Sentry event to traceback output.

Use os.path.join, not urlparse.urljoin

For failures, add http links to log directories.

Add a catch-all __getattr__(); add comments

Rewrite email-generating code.

Tweak logging

Merge pull request #52 from ceph/wip-6125

Don't hardcode path to teuthology-suite

internal: remove (hopefully obsolete) kern.log checks

Signed-off-by: Sage Weil <sage@inktank.com>

internal: unbreak syslog BUG checks and gz

This broke back in ace4cb07b2de99644c63f3ab90c21a663a384e69

Signed-off-by: Sage Weil <sage@inktank.com>

Don't hardcode path to teuthology-suite

Merge pull request #51 from ceph/wip-queue

queue: only git fetch once per minute per branch

ceph_manager: wait for dump_ops_in_flight on osd revival

Wait for a command that implies a complete startup instead of
'version' (which does not).

Fixes: #5924
Signed-off-by: Sage Weil <sage@inktank.com>

queue: only git fetch once per minute per branch

This takes 1-2 seconds and makes launching jobs slow.  Only do it once
every 60 seconds per branch.

Signed-off-by: Sage Weil <sage@inktank.com>

Merge pull request #50 from ceph/wip-worker-lock

queue: only let one worker update the teuthology checkouts at a time

queue: only let one worker update the teuthology checkouts at a time

Signed-off-by: Sage Weil <sage@inktank.com>

Fix for blank yaml.

As well as fix variabl variable naming.

Update call to get_sentry_client()

Use teuthology.config

Add teuthology.config, the start of a better system

Merge pull request #49 from ceph/sentry

Rewrite branch fetching.

Merge pull request #48 from ceph/wip-nuke-firmware

nuke: clean up stray firmware.git locks

nuke: clean up stray firmware.git locks

These get lost occasionally and cause all firmware.git updates to
fail when the kernel task runs.

Signed-off-by: Sage Weil <sage@inktank.com>

Move git stuff to fetch_teuthology_branch()

Rewrite branch fetching.

queue: fetch origin, not branch

Signed-off-by: Sage Weil <sage@inktank.com>

suite: only skip on non-vps if os-type is actually defined

If it's not defined, as it is (not) with most of the nightlies,
proceed!

Signed-off-by: Sage Weil <sage@inktank.com>

queue: only bootstrap new checkouts

Until we figure out why bootstrap is getting stuck like this:

 9851 pts/7    S      0:03 /home/teuthworker/teuthology-master/virtualenv/bin/python ./teuthology-master/virtualenv/bin/teuthology-worker -v --archive-dir /var/lib/teuthworker/archive --tube plana --log-dir /var/lib/teuthworker/archive/worker_logs
 2075 pts/7    Z      0:00  \_ [git] <defunct>
 2112 pts/7    Z      0:00  \_ [git] <defunct>
 2138 pts/7    Z      0:00  \_ [bootstrap] <defunct>
 9852 pts/7    S      0:03 /home/teuthworker/teuthology-master/virtualenv/bin/python ./teuthology-master/virtualenv/bin/teuthology-worker -v --archive-dir /var/lib/teuthworker/archive --tube plana --log-dir /var/lib/teuthworker/archive/worker_logs
 2153 pts/7    Z      0:00  \_ [git] <defunct>
 2177 pts/7    Z      0:00  \_ [git] <defunct>
 2264 pts/7    Z      0:00  \_ [bootstrap] <defunct>
 9853 pts/7    S      0:03 /home/teuthworker/teuthology-master/virtualenv/bin/python ./teuthology-master/virtualenv/bin/teuthology-worker -v --archive-dir /var/lib/teuthworker/archive --tube plana --log-dir /var/lib/teuthworker/archive/worker_logs
 2141 pts/7    Z      0:00  \_ [git] <defunct>
 2276 pts/7    Z      0:00  \_ [git] <defunct>
 2305 pts/7    Z      0:00  \_ [bootstrap] <defunct>
 9854 pts/7    S      0:03 /home/teuthworker/teuthology-master/virtualenv/bin/python ./teuthology-master/virtualenv/bin/teuthology-worker -v --archive-dir /var/lib/teuthworker/archive --tube plana --log-dir /var/lib/teuthworker/archive/worker_logs
 7448 pts/7    Z      0:00  \_ [git] <defunct>
 7449 pts/7    Z      0:00  \_ [git] <defunct>
 7450 pts/7    Z      0:00  \_ [bootstrap] <defunct>
 7452 pts/7    Z      0:00  \_ [teuthology-resu] <defunct>
 9855 pts/7    S      0:01 /home/teuthworker/teuthology-master/virtualenv/bin/python ./teuthology-master/virtualenv/bin/teuthology-worker -v --archive-dir /var/lib/teuthworker/archive --tube plana --log-dir /var/lib/teuthworker/archive/worker_logs
 7712 pts/7    Z      0:00  \_ [git] <defunct>
 7713 pts/7    Z      0:00  \_ [git] <defunct>
 7714 pts/7    Z      0:00  \_ [bootstrap] <defunct>
 7716 pts/7    Z      0:00  \_ [teuthology-resu] <defunct>
 9856 pts/7    S      0:03 /home/teuthworker/teuthology-master/virtualenv/bin/python ./teuthology-master/virtualenv/bin/teuthology-worker -v --archive-dir /var/lib/teuthworker/archive --tube plana --log-dir /var/lib/teuthworker/archive/worker_logs
 2316 pts/7    Z      0:00  \_ [bootstrap] <defunct>
 9857 pts/7    S      0:03 /home/teuthworker/teuthology-master/virtualenv/bin/python ./teuthology-master/virtualenv/bin/teuthology-worker -v --archive-dir /var/lib/teuthworker/archive --tube plana --log-dir /var/lib/teuthworker/archive/worker_logs
 2340 pts/7    Z      0:00  \_ [bootstrap] <defunct>
 9858 pts/7    S      0:01 /home/teuthworker/teuthology-master/virtualenv/bin/python ./teuthology-master/virtualenv/bin/teuthology-worker -v --archive-dir /var/lib/teuthworker/archive --tube plana --log-dir /var/lib/teuthworker/archive/worker_logs
23188 pts/7    Z      0:00  \_ [bootstrap] <defunct>
 9859 pts/7    S      0:03 /home/teuthworker/teuthology-master/virtualenv/bin/python ./teuthology-master/virtualenv/bin/teuthology-worker -v --archive-dir /var/lib/teuthworker/archive --tube plana --log-dir /var/lib/teuthworker/archive/worker_logs

Signed-off-by: Sage Weil <sage@inktank.com>

Do not run multiple tests (for distros) on baremetal.

Teuthology doesnt care about os_type for baremetal (ATM). This
change makes it so you can run tests that have been switched over
to run on multiple distros (on vms) on baremetal as well as all
non-ubuntu tests will be skiped (to avoid running the same test
multiple times on baremetal.

Signed-off-by: Sandon Van Ness <sandon@inktank.com>

Allow exclude_arch and exclude_os_type in qa-suite yamls

Adding the option:

exclude_arch:

or

exclude_os_type:

in the ceph-qa-suite yaml allows tests to be skipped for certain
types of hardware or distros.

Example:

exclude_arch: armv7l

This will make said test not run on arm machines.

exclude_os_type: rhel

Would make multi-distro tests skip a specific test on RHEL.

Fix SyntaxError

Merge pull request #47 from ceph/sentry

Make the worker obtain the correct teuthology branch

Worker shouldn't attempt to rebuild an existing virtualenv

Use the ceph.com git mirror.

Use fetch and reset --hard, not checkout and pull

Make the worker obtain the correct teuthology branch

Merge pull request #45 from ceph/wip-6084

Fix upgrading edge-ish case

Fix upgrading edge-ish case

install: use get() in _get_config_value_for_remote

Fixes: #6080
Signed-off-by: Sage Weil <sage@inktank.com>
Tested-by: Tamil Muthamizhan <tamil.muthamizhan@inktank.com>

Merge pull request #44 from ceph/sentry

Add initial support for sentry

radosgw-admin: use check_status for task success

Alter the code to use the 'check_status=True'
option in rgwadmin() rather than following the
call with 'assert not err'. Should make the
tests a bit more clear and result in a more
useful error (throw the call stack rather than
just 'assert new err failed').

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

radosgw-admin: adding radosgw-admin tests

Adding tests for syncing buckets between
RGWs. Fulfills ticket #5605.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Make client a global variable

Beginnings of support for Sentry.