v3.9.4 beta test

[HansR]

As announced in the Installation thread, there is a new version coming up with FTP locking so that the corruption of the stationswithutils.xml can be prevented and manual interaction from my side is no longer required. The test which I would like to see run before releasing it is described in the accompanying file. The executable is a plain fully functional, with no additional fancy logging , version of CumulusUtils so it can be applied directly into your Cumulus ecosystem. You could even use it before the test if you have confidence, it is a real beta though

The only requirement is the timing of the test (fri 3 July, 11h UTC). Further description in the accompanying text file.
Any comments, remarks or suggested/required modifications, please let me know.
The test is open for all CumulusUtils users of course.

[billy]

All set up here. There may be a little delay in getting the logs returned to you.

[HansR]

OK. Great!
Delay in sending the logs is better than delay in syncing the runs
(NOTE: you can start using the executable, it is a perfectly healthy one with just a new piece of code in need of verification. You never now what that may bring )

[kocher]

Tomorrow I will have everything ready.

The only obstacle is that my automatic cumulusUtils FTP doesn't work; that is, I carry out the transfer with other software that I will give 15 seconds of waiting after executing "cumulusUtils.exe mapson".

Is this correct? or should I reduce that space of time

[HansR]

Thanks Javier.

Aha, interesting situation.
Actually, for the test, I would say it is not required to actually upload the resulting maps.txt file (if automatic no need to bother of course) so I would say: don't do the upload.
Saves you the stress of manual timing.

Just start the command every minute.
The lockfile and the upload have no relation because the lockfile is created elsewhere.

Hope for the rest everything is clear?

[kocher]

OK perfect.
Everything is ready to run repeatedly from 11:00 (UTC), 13:00 CEST (local time).

Thank you very much for your interest

Saludos desde San Sebastian

[BeaumarisWX]

Hi Hans,
All setup in Tassie,
I will most likely be in the land of nod by 21:00 AEST tonight, but will send logs morning of the 4th.
kind Regards,

[HansR]

Great guys! We have 4 participants (12 downloads though ) that should be enough.
I really am curious if we are able to create at least one collision!
What I hoped did not occur in the old one, I wish for now

Log files a bit later is of course not a problem, I'll analyse over the weekend.

[kocher]

It seems that everything has worked as expected.

I attach the log files

[HansR]

OK, the test has finished just a few minutes ago and as far as I could see - I followed the creation and deletion of the lockfiles - it all went perfectly well.
I could see them following each other, and when there were two lockfiles one disappeared (failing the lock) so I expect the logfiles to well show these situations.

Thanks for preparing and executing the test!
If the logfiles show what I expect and the system functions as designed I expect to release Sunday or Monday. It gave me a good feeling.

Thanks again! I'll let you know how I get on with the results.


[EDIT:] @kocher: indeed it did work as expected

[BeaumarisWX]

Hi Hans,
3 of the 10 tests where successful and Map.txt seems fine.
Attached all test files, including last Map.txt which was at 21:06, also attached the normal 5 min run logs fyi, note file names change when exe running 2 instances.

maptest_tassie.zip

Kin Regards,

[HansR]

@Tony: Thanks! The running of the normal 5 min run is an unexpected complication (I had deleted my normal 10 min runs) but it is interesting to note that the complicated result of that was that it created locks itself and the situation was resolved perfectly. So I am happy with it.

No orphaned lockfiles remained after the test.
That file naming for duplicated instances is a nice plus

So: Thanks. Happy me.

[BeaumarisWX]

Hi Hans,
Yes all true, good result.
Thank you for arranging it, well done.
oh forgot to mention, the sysinfo mods work perfect too, thankyou.
Kind Regards,

[billy]

Hi Hans, Good result! herewith my logs.

[HansR]

Thank you Billy.

This actually makes it very clear we did create collisions (look at the time it sometimes take to get a lock!!), it was a very successful test. Furthermore I am happy we did not have more participants because that would have made it a lot more difficult to analyse and very probable that few would have been able to get a lock with the timespan.

It also made clear the timeout I implemented (5 seconds) is on the limit and sometimes was not enough.

So, I will make a somewhat longer timespan (8 seconds) and I will create some extra messaging to report the number retries.
Will release soon.

Thanks all