series/instances are not deleted from online/cache storage - instance availability

[zea...@gmail.com]

Dear all,

we have the problem that studies are not deleted from the online (cache storage) if parts of the study have already been received and transferred to the nearline storage (permanent). The instances or series which are received some days/hours later for this study are exported to the nearline storage via the exporter, but the instance availability of this series then remains ONLINE. The availability of the study, however, remains NEARLINE.

These individual series/instances are then not deleted from the online storage, which is why the Online Storage threatens to fill up.

We use version 5.23.2.

[Vrinda Nayak]

Have you configured any Deleter Thresholds and specified Export Storage ID as your nearline storage (in storage descriptor configuration for online storage) ?

Please share your storage configurations.

[zea...@gmail.com]

Hi Vrinda,

thanks for your answer. Yes we configured a  Deleter Thresholds and specified an Export Storage ID .

[Vrinda Nayak]

I've used exactly same configuration as yours (except deleter threshold value), the studies do get deleted. Attached my fs1 (ONLINE CACHE) and nearline (NEARLINE PERMANENT) storage configurations. Also, do check if there is any value configured for Purge Storage Polling Interval.

[zea...@gmail.com]

Hello Vrinda,

thank you very much.This also works for us with most of the studies, but there are still some studies that have not been deleted, even though all the criteria apply and are also stored in the nearline and that is why the online storage  threatens to fill up, because these studies are not deleted.

 I have had a look in the database.

pacsdb=> select count(l1.pk) from location l1 where l1.storage_id='online' and exists (select l2.pk from location l2 where l2.storage_id='nearline' and l1.storage_path=l2.storage_path);

 count

--------

 660475

(1 row)

pacsdb=> select count(l.pk) from location l where l.storage_id='online';

  count

----------

 660475

(1 row)

pacsdb=> select count(l.pk) from location l where l.storage_id='nearline';

  count

----------

 29510471

(1 row)

The strange thing is that it affects studies that were transmitted between 6 July and 19July. Attached there ist one study where series 1 was received on 16 of July and series 2 was received on 19  of July. Study availability is nearline, series 1 availability is also nearline but series 2 availability is online.

[Vrinda Nayak]

> The strange thing is that it affects studies that were transmitted between 6 July and 19July. Attached there ist one study where series 1 was received on 16 of July and series 2 was received on 19  of July. Study availability is nearline, series 1 availability is also nearline but series 2 availability is online.

Check if Series 2 is exported to nearline or not. For test purpose, manually trigger export of series 2 to nearline (queued exporter) and wait for purge storage interval to verify that study containing 2 series which are now both available on nearline gets deleted or not.

[zea...@gmail.com]

We have tried this several times, but unfortunately the series does not get deleted. 

Attached are the log messages.  

[Vrinda Nayak]

Not reproducible - I tested with some more sample studies / series.

Your log however does show

2021-08-18 16:03:15,443 INFO  [org.dcm4chee.arc.delete.impl.PurgeStorageScheduler] (EE-ManagedExecutorService-default-Thread-41042) Usable Space on CACHE Storage[id=online, uri=file:///storage/online/] below 30GB - start deleting 17.341459456GB

post export to nearline storage task completion.

Have you checked on the filesystem or in DB for deletion of series (exported to nearline)?

[zea...@gmail.com]

Hello Vrinda, yes we have checked that. All instances that are still on the online filesystem are also available on the nearline filesystem  and have two entries in the location table of the DB. Unfortunately, the studies will not be deleted in the period from 6 July to 19 July. However, new received studies are processed normally and are getting deleted.

[zea...@gmail.com]

hi vrinda we have exactly the same problem now with another installation. 

The  migration is done and we have started a second run of data which failed in the first run. Then, of course, some series of studies, which were already transfered , were added. Now these are not recognised by the archive as deletable because the study is already set to 'nearline'. 

vrinda...@j4care.com schrieb am Dienstag, 24. August 2021 um 19:35:30 UTC+2:

[zea...@gmail.com]

the wildfly/tmp directory is full of tmp files from the days we started the second round of migration. does this have something to do with it?

[zea...@gmail.com]

Hi Vrinda, thanks for your help.

Solved it. After upgrading to Version 5.24.1 it works. 

I think it had something to do with this issues.

https://github.com/dcm4che/dcm4chee-arc-light/issues/3139

Export of studies available only on nearline storage throws NullPointerException

https://github.com/dcm4che/dcm4chee-arc-light/issues/3026

Update instance availability for instances on retrieve of studies from nearline storage to online/cache storage
Update instance availability for instances present on nearline storages on deleting studies from online/cache storage

https://github.com/dcm4che/dcm4chee-arc-light/issues/3008