CVS Completely Broken for Our Builds in 1481

I upgraded to 1481 this morning.

- Stopped the server and agents.
- Unpacked the new Team City.
- Reapplied changes for database usage (including pointing to an new, empty database).
- Removed work directories on build agents.
- Moved the .BuildAgent directory out of the way and copied the config directory to a new .BuildAgent (removing the properties files in configuration subdirectories).
- Restarted the server.
- Restarted the agents.

So not a complete fresh start, but pretty close. I think all I brought over were the project-config.xml files, a few other configuration files and the buildAgents to be upgraded (the Default Agent was fresh).

And every single build broke. I should say, things didn't even get to the build. I get the error:

Error while applying patch: malformed input around byte 0

For every build. XStream, which I kept around from the default config, is not having this problem, but it is still failing. Here is some stuff from the agent:

WARN - rver.agent.impl.BuildAgentImpl - java.io.UTFDataFormatEx
ception: malformed input around byte 0
java.io.UTFDataFormatException: malformed input around byte 0
at java.io.DataInputStream.readUTF(DataInputStream.java:612)
at java.io.DataInputStream.readUTF(DataInputStream.java:522)
at jetbrains.buildServer.vcs.patches.AbstractPatcher.applyPatch(Abstract
Patcher.java:45)
at jetbrains.buildServer.agent.impl.BuildAgentImpl.applyPatch(BuildAgent
Impl.java:463)
at jetbrains.buildServer.agent.impl.BuildAgentImpl.access$400(BuildAgent
Impl.java:46)
at jetbrains.buildServer.agent.impl.BuildAgentImpl$2.run(BuildAgentImpl.
java:343)
at java.lang.Thread.run(Thread.java:595)
INFO - rver.agent.impl.BuildAgentImpl - Patch applied in 185 se
conds
INFO - rver.agent.impl.BuildAgentImpl - Found runner Ant for

And from the server:

INFO - serverSide.impl.DBBuildHistory - Start creating history
entry for Compile 15, date: 1153846321000
INFO - serverSide.impl.DBBuildHistory - End creating history
entry for Compile 15
can't parse conflict ' Dec 19 00:00:00 0001'
INFO - erverSide.impl.BuildServerImpl - Running build saved to
DB: 17 Compile
can't parse conflict ' Dec 31 00:00:00 0001'
WARN - .serverSide.impl.BuildTypeImpl -
java.lang.NumberFormatException: For input string: ""
at java.lang.NumberFormatException.forInputString(NumberFormatException.java
:48)
at java.lang.Long.parseLong(Long.java:424)
at java.lang.Long.parseLong(Long.java:461)
at java.text.DigitList.getLong(DigitList.java:167)
at java.text.DecimalFormat.parse(DecimalFormat.java:1271)
at java.text.SimpleDateFormat.subParse(SimpleDateFormat.java:1692)
at java.text.SimpleDateFormat.parse(SimpleDateFormat.java:1124)
at java.text.DateFormat.parse(DateFormat.java:333)
at org.netbeans.lib.cvsclient.response.ResponseUtils.parseDateString(Respons
eUtils.java:30)
at org.netbeans.lib.cvsclient.response.ResponseParser.processResponse(Respon
seParser.java:171)
at org.netbeans.lib.cvsclient.RequestProcessor.handleResponses(RequestProces
sor.java:272)
at org.netbeans.lib.cvsclient.RequestProcessor.processRequests(RequestProces
sor.java:182)
at org.netbeans.lib.cvsclient.RequestProcessor.processRequests(RequestProces
sor.java:84)
at org.netbeans.lib.cvsclient.command.checkout.ExportCommand.checkout(Export
Command.java:186)
at org.netbeans.lib.cvsclient.command.checkout.ExportCommand.execute(ExportC
ommand.java:76)
at jetbrains.buildServer.buildTriggers.vcs.cvs.CvsCommandExecutor.executeCom
mand(CvsCommandExecutor.java:12)
at jetbrains.buildServer.buildTriggers.vcs.cvs.CvsPatchProvider.doExport(Cvs
PatchProvider.java:33)
at jetbrains.buildServer.buildTriggers.vcs.cvs.CvsPatchProvider.processExpor
t(CvsPatchProvider.java:5)
@

I tried restarting a buildAgent to make sure that there wasn't something in the update process which caused a problem. No luck.

I created a new project in case the project config was odd. No luck.

It looks like things are dying in the same general place in terms of the patch. I'm guessing there is something in our CVS repository which is causing whatever TC uses for CVS to barf. Or something like that.

It's a real shame as it looked like a pretty nice build.

I'm going to do a little more testing and prodding to see if I can resolve this. If not, I'll put all the details into a Jira issue.

--Tim

5 comments
Comment actions Permalink

Tim McNerney wrote:

I upgraded to 1481 this morning.

>

And every single build broke. I should say, things didn't even get to the build. I get the error:

Error while applying patch: malformed input around byte 0

For every build. XStream, which I kept around from the default config, is not having this problem, but it is still failing. Here is some stuff from the agent:

WARN - rver.agent.impl.BuildAgentImpl - java.io.UTFDataFormatEx
ception: malformed input around byte 0
java.io.UTFDataFormatException: malformed input around byte 0


I'm seeing the same problem, but only with the Build Agent that's running on a Linux box:

http://www.jetbrains.net/jira/browse/TW-603

I think it has trouble to check out binary files.

Sascha

0
Comment actions Permalink

I'm seeing the same problem, but only with the Build
Agent that's running on a Linux box:

http://www.jetbrains.net/jira/browse/TW-603

I think it has trouble to check out binary files.


Interesting. Our agents are on Linux. I'll try with a Windows agent and see if I get the same.

--Tim

0
Comment actions Permalink

Just tried and I'm running into the problem with an agent running on Windows. The server is still on Linux and I'm doing the checkout on the server, so it may just be the checkout.

--Tim

0
Comment actions Permalink

Tim,

It looks like we found out the reason of the problem (nasty shared buffer). Moreover, we faced
the same problem here for our own builds (or at least it looks the same). We plan to issue beta of
TeamCity today and I really hope this problem will be fixed in it.

Thanks for your help and patience.

Kind regards,
KIR

Tim McNerney wrote:

Just tried and I'm running into the problem with an agent running on Windows. The server is still on Linux and I'm doing the checkout on the server, so it may just be the checkout.

--Tim



--
Kirill Maximov
Software Developer
JetBrains, Inc.
http://www.jetbrains.com
"Develop with pleasure!"

0
Comment actions Permalink

Awesome.

Thanks.

--Tim

0

Please sign in to leave a comment.