Disable Cluster log

[jduarte12]

Hello,

Last friday, we have upgraded from Qlikview Server 11.2 SR8 to 11.2SR9 in order to solve a known bug of the previous version.

However, after the weekend, we noticed that somehow the available disk space in C:\ partition had decreased significantly. In fact, we were in danger of running out of space. Then, we looked for possible sources for that and discovered in

C:\ProgramData\QlikTech\DistributionService\1\Log

one Cluster_yyyymmdd.txt file per day since the upgrade to SR9. Each one of these files was larger than 1Gb, which means that after deleting these files we had the typical values of available disk space back.

Meanwhile, we were searching in the management console where could I change the definition for disabling (or simply getting less information) these Cluster log files.

We recall that this space 'consumption' had never occured in the previous versions of Qlikview that we have had installed; thus, we believe that there will be a way for avoiding about 10Gb of Cluster log files in less than a week.

I would like your advice on what to do to correct that.

Thank you & Best regards,

João Duarte

[jduarte12]

Hello

After some extra research, I discovered that if I check "no logging" under System\Setup\Distribution Services\General\Settings for QDS, QlikView stops writing data in Cluster_yyyymmdd.txt files.

However, by doing that, we also loose the ability to get the other log files that we have always had and that refer to triggered tasks.

I don't know if it helps, but these Cluster log files contain the following message:

Failed to read from C:\ProgramData\QlikTech\DistributionService\TaskResults\TaskResult_[...].xml. Exception=System.FormatException: String was not recognized as a valid DateTime. ||    at System.DateTime.Parse(String s) ||    at QDSMain.TaskResult.ReadXml(XmlReader reader) ||    at QDSMain.Clustering.ClusterFileHandler.Read(Action`1 readXML)

Anyone has an idea on what can be the meaning of that? We would appreciate your help on this issue, as it is not wanted to get these huge cluster log files, but we want to keep the log files that we have always had.

Thank you & Best regards,

João Duarte

[jerrysvensson]

Have you contacted Qlik support?

We are just about to update to SR9. I find this alarming.

[jduarte12]

Jerry,

I could not contact QlikSupport yet. Nevertheless, if I don't get any feedback from the community, I will certainly do so in the beginning of the next week.

[jfkinspari]

Have you checked the content of the Cluster log file?

It might contain millions of records of the same error, so instead of removing the logfile you should solve the issue giving the error.

[jduarte12]

Jens,

That is in fact what is happening.

Perhaps I haven't explained it properly, but we are interested in solving it, either by managing to disable the creation of cluster log files (and only those) or by correcting whatever is causing those millions of records of the same error. Removing periodically the log file is the sole quick solution that we can apply in order to keep some disk space until we solve this issue.

Still, when we used either 11.0SR1 or 11.2 SR8, this problem did not occur. Is there any parameterization that is specific to this latest service release?

[jfkinspari]

Okay, I am not aware of any options disabling only the Cluster log-file. However, it's probably somewhere...

What does the error say? It might hint to finding a solution.

[jduarte12]

Yes. I have tried to find a disable option, but I haven't found anything yet.

Below you will find a line of the log file:

28-Nov-14 10:25:47.4368256	Error	Failed to read from C:\ProgramData\QlikTech\DistributionService\TaskResults\TaskResult_0188f054-943a-4f98-8620-8f4c758e799f.xml. Exception=System.FormatException: String was not recognized as a valid DateTime. ||    at System.DateTime.Parse(String s) ||    at QDSMain.TaskResult.ReadXml(XmlReader reader) ||    at QDSMain.Clustering.ClusterFileHandler.Read(Action`1 readXML)

There is a line for each timestamp entry.

[jfkinspari · Accepted solution]

Okay, I can see that my own Cluster log file also have had a bunch of these (66, 63 mb log files) but then stops.

Is it always the same result XML file the error references?

I would open the XML file, and find out which reload job is referenced in the xml, and see if that job might be seek!

Perhaps delete and recreate the job.

Just in case you haven't, you should try and restart the QlikView services.

[jduarte12]

Jens,

First of all, sorry for the delay in my answer. In fact, I have done as you suggested and noticed that the cluster log file referred always to the same 6 xml. So, I discovered that each one of these xml corresponded to tasks that were disabled in the Qlikview Management Console. As these tasks are not needed to perform (whereby they were disabled), I simply cancelled them at all. After that, the cluster log file for today stopped growing.

So, I think I may conclude that there is something in 11.2SR9 that does not cope well with disabled tasks in the QMC, which makes arise the issue I reported in my initial post.

Thank you very much for your help.

João Duarte