GSB Forums

Forum and cloud outage

admin - 10-12-2018 at 03:45 PM

At roughly 3pm central usa time Tue 11 December 2018 , the forum, trademaid.info web site and GSB cloud will go down for < 1hour. If its longer than this it means something has gone wrong. That being the case peterzwag@gmail.com will be best means of communication. This outage to to cope with growth of GSB client base to give more resources to the cloud.

admin - 11-12-2018 at 08:43 PM

upgrade complete :)
Only took a few minutes

admin - 9-8-2022 at 01:48 AM

There is an outage on the GSBcloud. Standalone will work ok. Cloud workers are almost all not working, and build of systems (IE 20,000) completes way too early. Looking into it now

admin - 9-8-2022 at 05:37 AM

Apologizes, cloud is still down. We are still working on it. Not easy to diagnose this.

admin - 9-8-2022 at 06:51 AM

problem seems to be intermitent. Some machines work fine, others dont. I stop work for the day now

admin - 9-8-2022 at 07:16 AM

would like to know who has issues? many of my machines work fine, others dont. Problem is few or no workers, even when workers are free. Today we did find a data base issue where not all workers are used - but this has been an issue for a long time. It will be resolved as time permits. All Manager jobs were deleted in the data base. (sorry but this was one of the things we tried to fix the issue)
Problem can also be workers not reporting graph, but are working.
Other big issue is manager says its completed before the 20,000 systems or so are made.

TheMetalDog - 9-8-2022 at 09:52 PM

Quote: Originally posted by admin  
would like to know who has issues? many of my machines work fine, others dont. Problem is few or no workers, even when workers are free. Today we did find a data base issue where not all workers are used - but this has been an issue for a long time. It will be resolved as time permits. All Manager jobs were deleted in the data base. (sorry but this was one of the things we tried to fix the issue)
Problem can also be workers not reporting graph, but are working.
Other big issue is manager says its completed before the 20,000 systems or so are made.


Hi Peter,

I had some local workers that were not starting.

The run this was experienced had a manager set to load a data file to be from 06/01/2016 - 07/31/2022, 1 minute with tick vol, 00:00 - 23:59. Is there a chance that the large data file hit some computers in the cloud causing issues because of the size of the data file? It was around 295MB







admin - 9-8-2022 at 10:10 PM

@themetaldog.
If you have 1 min data, x 30 to make 30 minute, the file uploaded would be 30 minute = 10 mb
It could take considerable time to upload and download 300 mb if it was 1 minute.
What did you use?

Apart from myself and possibly one other user, no one else seems to have an issue

TheMetalDog - 9-8-2022 at 10:40 PM

Quote: Originally posted by admin  
@themetaldog.
If you have 1 min data, x 30 to make 30 minute, the file uploaded would be 30 minute = 10 mb
It could take considerable time to upload and download 300 mb if it was 1 minute.
What did you use?

Apart from myself and possibly one other user, no one else seems to have an issue


@admin

2 timeframes, 15 and 30 minute bars.

It sounds like the app doesn't upload the original data but compresses it into the rolled up bar time? Does it also cut the dates off to what will be used for that run? e.g. if the date range on the original dataset contains 10 years worth of data, but Manager is set to only use 5 of it for the current run, does only 5 years of rolled up data get sent to the Workers?

That would be ideal for efficiency, as well as convenient. I like the idea of having one file of one-minute bars covering all the years of data needed, and then setting the session time and interval in the Manager to squeeze just what is needed from that data file.


admin - 9-8-2022 at 11:58 PM

I think the files are zipped up when sent to workers. Not sure on what happens with dates if they are reduced.
Understand why you like one data set good for all.
Im hard pressed to ever beat 30 min bars. 15 30 is worth trying but not likely to be better than 30 min
it will also use about 3 times more ram than just 30 min

admin - 10-8-2022 at 04:19 AM

Likely the outage is fixed. Only affected a few people. Cause was time zone somehow changed on server. That shouldn't have mattered as time zone in GSB code is converted to utc
We did find some other issue to be fixed later, that will cause the cloud to be used more effectively.

Daniel UK1 - 10-8-2022 at 06:13 PM

Today and (yesterday I think) the cloud has been down most of the day, a manager starts, gets a few systems built and then looses all workers, and then it stays in "searching" for hours..
is there a known issue with the cloud today ? there is plenty of people sharing workers, so i should have no problem getting assigned workers, but nada..

Any ideas whats wrong ?
I also seem to be getting lots of exception errors showing etc..

admin - 10-8-2022 at 07:39 PM

Yes, all should be fixed. You may have to restart managers and possible workers.
Basically time wrong on gsb data base server caused all the issues. Apologies. It wasted us all a lot of time

admin - 20-11-2024 at 08:28 AM

critical alert. I have 3 ts data feeds. all have no data for last 3/4 hours. Any one else? ts phones I cant get to work, apart from trade desk. on hold now

admin - 20-11-2024 at 08:33 AM

ts back on line. was down for 45 min