Replication data is corrupted

[ljp]

I am working remotely with a Windows VPN machine and a separate Linux box which I use for development.

 

The VPN links to a Premium Plastic SCM installation which works well (but slowly over the internet). To speed things up and to allow me to make full use of Plastic SCM features on my development machine I wish to create replication packages on the windows box and transfer them to (and from) the Linux box (which currently has a community edition Plastic SCM installed) via USB stick.

 

When I try to create the initial package files on the windows box, either by command line (e.g. cm replicate br:/main/2014.1_dev@global@plastic:8087 --package=2014.1.pk) or through the GUI I get one of the following 2 errors:

 

Replication data is corrupted and the operation can't be completed.
Specific Error: 'Too many bytes in what should have been a 7 bit encoded Int32'

or for a smaller branch:

Replication data is corrupted and the operation can't be completed.
Specific Error: 'Unable to read beyond the end of the stream.'

 

I cannot find any reference to these messages on the Plastic site or forums. Does anyone have any idea what these messages mean and how to fix them?

[manu]

Hello!

 

I guess you have two Plastic SCM servers right? The one in Linnux and the one in Windows.

Can you plase give us the release you are using? I also need it for the "cm" client installations. (cm version)

 

I'll test the same just int case...

 

Thanks

[ljp]

All servers and clients are running 5.0.44.534

 

Thanks

[manu]

Can you repeat the replicate command with the "--stack" options:

 

cm replicate br:/main/2014.1_dev@global@plastic:8087 --package=2014.1.pk --stack

 

command and send us the output and the server log?

 

Seems to be a serialization error :S

[ljp]

The error message when running the command is:

 

C:\Users\ljp>cm replicate br:/main/2014.1_dev@global_admin@rs-plastic:8087 --package=2014.1.pk --stack
FetchingRevisions. 597 objects
FetchingReferences. 3614 objects
FetchingFinished
Replication data is corrupted and the operation can't be completed. Specific error: 'Too many bytes in what should have
been a 7 bit encoded Int32.'.
   at Codice.CM.Common.Serialization.Replication.BasePackageReader.ReadRevInfosAndChildren(IList& children)
   at Codice.CM.Common.Serialization.Replication.BasePackageReader.GetRevisions()
   at alf.a(String A_0)
   at fr.a(RepositoryInfo A_0, Guid A_1, Int64 A_2, Int64 A_3, Boolean A_4, String A_5)
   at c9.a(RepositoryInfo A_0, Guid A_1, Int64 A_2, Int64 A_3, Boolean A_4, String A_5)
   at c9.a(b A_0)
   at w1.c(String[] A_0)

 

I have got in touch with our administrator to get hold of the server log. I'll upload it when I can get hold of it.

 

Thanks
 

[ljp]

The administrator has asked me for the path name of the server log file you need.

[manu]

Sure, the file name is plastic.server.log and you will find it at "c:\programfiles\plasticscm\server".

[ljp]

Log file as requested. LJP_Plastic_server_log.txt

[manu]

Sorry I can't see the error trace inside the file. Can you please enable the client (cm) log and send us the result? http://www.plasticscm.com/infocenter/technical-articles/kb-enabling-logging-for-plastic-scm-part-i.aspx

 

Can you try to recreate the package? Was the package copied? In that case can you check the checksum at the source and destination location?

 

I guess you can't share with us the file.... We can sight an NDA agreement so then we can fix it faster if the issue it with the package....

[ljp]

I have managed to get a client log file cm.log.txt

 

I do end up with a very small package file of 2.5Mb locally. Should there be a duplicate of it on the server? If so, where.

 

The GUI goes through:

 

fetching revisions (~900)

fetching branches

fetching links

fetching references (~4000)

fetching SEIDs (0)

fetching data - fails