NotificationRegistryInterface.getProgress

[halmsx]

hi

sometimes a checkin failed. this can cause by many reasons, leaving the topicid progress result empty using NotificationRegistryInterface.getProgress

{ "response": { "result": { } } }

when this happened, i cant do any new checkin. for now i have to restart the bimserver.

question, is there any way for remove this 'topic' using the api and remove the 'blocking' without having to restart the server?

thanks.

[rubendel]

Not sure what this question has to do with the NotificationRegistryInterface.getProgress call.

There is probably a bug somewhere that causes the Project.checkinInProgress field not to be cleared. Can you give more details on what has happened when it doesn't get cleared?

[halmsx]

for me to reproduce this problem...

1. a user create a project and checkin a model.
2. while the checkin progress still running,
3. another user create another project and checkin another model

this doesnt happens always. but half the time it does.

i run BIMServer on tomcat with XMS 512MB and XMX 4096GB and the ifc files are between 20-100MB each.

[rubendel]

4096GB is a bit much ;) You probably mean 4096MB, which could easily be too little. Are there any errors in the log file when this happens?

[TitouanGisle]

Hello ! I frequently have the same problem, or similar : whenever a checkin fail, the associated topic remains, and I am unable to checkin new files.

The logs don't display anything about it. The API returns the message "Checkin in progress on this project (topicId: XXX), please try again later" whenever I try to checkin a new file. The "getProgress" method returns an empty result. The method "cleanupLongAction" returns an empty result, and the topic remains. Reseting the server does delete the topics. I am on version 1.5.162.

It is to be noted that I used the methods initiateCheckin->checkinInitiatedAsync. I tried today with checkinAsync, and I don't seem to have this problem.

[halmsx]

it seemed that if there is an error during checkin, the topic id remains in the que.

if i tried to checkin another file afterward, id get the following...

{'response': {'exception': {'__type': 'UserException', 'message': 'Checkin in progress on this project (topicId: 11), please try again later'}}}

if i run NotificationRegistryInterface.getProgressTopicsOnServer i can see topic id 11 in the list. but NotificationRegistryInterface.getProgress will return empty.

i have to restart the bimserver to be able to checkin another file

[hlg]

I have experienced this issue too, but in any case this is preceded by a server error when the checkin fails. Examples situations are:

• URL checkin with invalid domain (e.g. typo or URL with query part)
• URL checkin with no deserializer selected
• broken pipe, interrupted connection

Probably these should be handled on the server side otherwise some failures (e.g. the last) can leave a mess behind. But taking that risk, Is there a way to clear a topicId without restarting the server?

[DaLiangLee]

same problem