Task #13130
closed
DataMiner - Update service on preprod to latest release 4.13.1
100%
Description
In order to work with DataMiner in Preprod it is necessary to update the service to the latest version gCube 4.13.1
Please check that both:
- dataminer1-pre.d4science.org
- dataminer-ghost-t.pre.d4science.org
are updated.
Updated by Roberto Cirillo over 6 years ago
- Target version changed from DataMiner - Provisioning to D4Science Infrastructure upgrade to gCube 4.13.1
Updated by Roberto Cirillo over 6 years ago
- Status changed from New to In Progress
The dataminer service is failing on gCube 4.13.1.
The service will be upgraded when the new version is available in 4.13.1
Updated by Roberto Cirillo over 6 years ago
- Status changed from In Progress to Paused
Updated by Roberto Cirillo over 6 years ago
- Status changed from Paused to In Progress
Updated by Roberto Cirillo over 6 years ago
- Status changed from In Progress to Feedback
- Assignee changed from Roberto Cirillo to Giancarlo Panichi
- % Done changed from 0 to 100
the dataminer instances have been upgraded.
Updated by Giancarlo Panichi over 6 years ago
- Status changed from Feedback to In Progress
- Assignee changed from Giancarlo Panichi to Roberto Cirillo
Hi @roberto.cirillo@isti.cnr.it , the service does not work there are dependencies that have not been correctly inherited (for example common-clients).
An update of the boms may be necessary.
Better to test smaller components like the uri-resolver first to see if they work.
Updated by Roberto Cirillo over 6 years ago
- Status changed from In Progress to Paused
I'm waiting a new build
Updated by Roberto Cirillo over 6 years ago
- Status changed from Paused to In Progress
- % Done changed from 100 to 90
New build available. I'm going to upgrade the service instances in preproduction
Updated by Roberto Cirillo over 6 years ago
- Status changed from In Progress to Feedback
- Assignee changed from Roberto Cirillo to Giancarlo Panichi
- % Done changed from 90 to 100
Updated by Roberto Cirillo over 6 years ago
- Status changed from Feedback to In Progress
- Assignee changed from Giancarlo Panichi to Roberto Cirillo
Updated by Roberto Cirillo over 6 years ago
- Status changed from In Progress to Feedback
- Assignee changed from Roberto Cirillo to Giancarlo Panichi
please @g.panichi@isti.cnr.it let me know if it works properly now
Updated by Giancarlo Panichi over 6 years ago
- Status changed from Feedback to In Progress
- Assignee changed from Giancarlo Panichi to Roberto Cirillo
- % Done changed from 100 to 0
@roberto.cirillo@isti.cnr.it there is a problem of conflicting versions on jersey libraries.
We made a fix and re-launched the build on ETICS.
Updated by Roberto Cirillo over 6 years ago
- % Done changed from 0 to 100
After a new deploy it works now.
Updated by Roberto Cirillo over 6 years ago
- Status changed from In Progress to Closed
Updated by Giancarlo Panichi over 6 years ago
- Status changed from Closed to In Progress
- % Done changed from 100 to 0
@roberto.cirillo@isti.cnr.it in pre-production there is the DataMiner service that I had fixed yesterday, it should still put what it has built in gCube 4.13.1 to check that the dependencies are now correct.
Please, check that the deletion of the file already_installed_algorithms.txt is added to the ansible script at the end of the installation
dataminer1-pre:~/wps_algorithms_install_log$ rm already_installed_algorithms.txt
In this way, when the dataminer is restarted loads all the algorithms correctly.
Updated by Roberto Cirillo over 6 years ago
Giancarlo Panichi wrote:
@roberto.cirillo@isti.cnr.it in pre-production there is the DataMiner service that I had fixed yesterday, it should still put what it has built in gCube 4.13.1 to check that the dependencies are now correct.
Please, check that the deletion of the file already_installed_algorithms.txt is added to the ansible script at the end of the installation
dataminer1-pre:~/wps_algorithms_install_log$ rm already_installed_algorithms.txtIn this way, when the dataminer is restarted loads all the algorithms correctly.
@g.panichi@isti.cnr.it the file "already_installed_algorithms.txt" should be already removed at the end of the update as reported here: #11578
Updated by Roberto Cirillo over 6 years ago
I'm going to reinstall the dataminer in preproduction
Updated by Roberto Cirillo over 6 years ago
- Status changed from In Progress to Feedback
- Assignee changed from Roberto Cirillo to Giancarlo Panichi
- % Done changed from 0 to 100
Done.
Updated by Giancarlo Panichi over 6 years ago
- Status changed from Feedback to In Progress
- Assignee changed from Giancarlo Panichi to Roberto Cirillo
@roberto.cirillo@isti.cnr.it , now the service has the right libraries, the Etics build is correct.
As for the file already_installed_algorithms.txt, however, it seems that the moment when the cancellation is done is not good.
Because if you check the file it's empty and the DataMiner service has not loaded any algorithms.
Perhaps the deletion should be done after restarting the service in the ansible script.
Updated by Roberto Cirillo over 6 years ago
Giancarlo Panichi wrote:
@roberto.cirillo@isti.cnr.it , now the service has the right libraries, the Etics build is correct.
Good news
As for the file already_installed_algorithms.txt, however, it seems that the moment when the cancellation is done is not good.
Because if you check the file it's empty and the DataMiner service has not loaded any algorithms.
Perhaps the deletion should be done after restarting the service in the ansible script.
Sorry but I don't understand what you are saying.
Yes, now the cancellation has been done before the restart of the smartgears container and the file now (after the container restart) is still empty. This means that at start time the file was empty.
If I will postpone the cancellation of the file after the contaienr restart what should change?
Updated by Giancarlo Panichi over 6 years ago
Change that the container does not reload the algorithms correctly.
Updated by Roberto Cirillo over 6 years ago
- Status changed from In Progress to Feedback
- Assignee changed from Roberto Cirillo to Andrea Dell'Amico
This is not an explanation and I didn't realize that. IMHO this is not a clean behavior and should be avoided.
@andrea.dellamico@isti.cnr.it do you think it is possible to do it by ansible since the container restart is managed as handler event?
Updated by Andrea Dell'Amico over 6 years ago
- Assignee changed from Andrea Dell'Amico to Roberto Cirillo
Roberto Cirillo wrote:
This is not an explanation and I didn't realize that. IMHO this is not a clean behavior and should be avoided.
@andrea.dellamico@isti.cnr.it do you think it is possible to do it by ansible since the container restart is managed as handler event?
It's doable but very complex. Also, that behaviour didn't change in many months: why is it a problem now?
Last: after an upgrade, all the algorithms are installed again.
Updated by Giancarlo Panichi over 6 years ago
- Status changed from Feedback to In Progress
Ok it's not a big deal for now.
Updated by Roberto Cirillo over 6 years ago
- Status changed from In Progress to Closed
