EOS:167: 1+0 configuration: testing and debugging

rajanikantchirmade commented 5 years ago

Write ST to check 1+0 cluster configuration and bootstrap
Verify IO using clovis apps (c0cp & c0cat)

Closes EOS-167

vvv commented 5 years ago

Nice work, Rajanikant! :thumbsup_tone1:

I have a couple of questions, which I will ask in a separate MR.

andriytk commented 5 years ago

reopened

andriytk commented 5 years ago

closed

typeundefined commented 5 years ago

I can see the updated comment in the source code. Thanks!

rajanikantchirmade commented 5 years ago

Yes, it is working. Please check output attached. eos-167-test.log.rtf

rajanikantchirmade commented 5 years ago

Returned $1 and not $?. check_io_return() clean up files and return last error passed to it.

rajanikantchirmade commented 5 years ago

Yes, removed.

rajanikantchirmade commented 5 years ago

Using two clovis apps, which required local endpoints.

If passed same endpoint to next clovis app it failing to in RC. I think that endpoint still in use and hence passing endpoint of next clovis app provisioned.

rajanikantchirmade commented 5 years ago

Done. Thank you.

rajanikantchirmade commented 5 years ago

added 1 commit

aba1f8dc - EOS-167: 1+0 configuration: testing and debugging

Compare with previous version

rajanikantchirmade commented 5 years ago

changed this line in version 9 of the diff

vvv commented 5 years ago

@rajanikant.chirmade Does this ST actually work?

vvv commented 5 years ago

What exactly is “checked” by this function? rm -rf never fails. And if it did fail, return statement would not be reached, because of set -e (set by st/common, which is sourced at line 6).

vvv commented 5 years ago

Not needed — H0_SRC_DIR and M0_SRC_DIR are set by h0. Please use h0 run-st t_1plus0.start-stop to run this script.

rajanikantchirmade commented 5 years ago

added 1 commit

c05c7358 - EOS-167: 1+0 configuration: testing and debugging

Compare with previous version

typeundefined commented 5 years ago

By the results of discussion with Max Medved I'll mark this comment as invalid. We do need the file to be random to be sure that the file was properly stored and copied by mero.

typeundefined commented 5 years ago

added 1 commit

20c17472 - Some bash script polishment

Compare with previous version

typeundefined commented 5 years ago

changed this line in version 7 of the diff

typeundefined commented 5 years ago

Can you please elaborate what should actually be fixed?

While reading the comment I don't quite understand whether it is a description what you are addressing or what you are working around. Since I'm probably not the only one who'll get confused here, maybe it's worthwhile to add clarity.

typeundefined commented 5 years ago

Reading from /dev/urandom can bring significant performance impact. Do you really need the content be of random garbage bytes? Will /dev/zero work here?

typeundefined commented 5 years ago

I would suggest to correct next three lines as follows:

LOG_FILE="$(mktemp --suffix _log_file)"
SRC_OBJ="$(mktemp --suffix _src_file)"
DEST_OBJ="$(mktemp --suffix _dest_file)"

The reason is that, in "$(mktemp)_log_file" expression, the file created by $(mktemp) will never be removed in your test. Moreover, mktemp guarantees that the file it returns will exist (while file "$(mktemp)_log_file" will obviously not).

rajanikantchirmade commented 5 years ago

added 1 commit