Storage failed: '>' not supported between instances of 'float' and 'NoneType'

[sylvestre]

Describe the bug codechecker store fails with

[INFO 2020-05-23 13:27] - Storage failed: '>' not supported between instances of 'float' and 'NoneType'

CodeChecker version

CodeChecker analyzer version:
---------------------------------------------------------------
Kind                 | Version                                 
---------------------------------------------------------------
Base package version | 6.13.0                                  
Package build date   | 2020-05-23T10:49                        
Git commit ID (hash) | 2934ac134ce141f7cc91a948cbba8e6389381fac
Git tag information  | 6.13                                    
---------------------------------------------------------------

CodeChecker web version:
------------------------------------------------------------------------
Kind                          | Version                                 
------------------------------------------------------------------------
Base package version          | 6.13.0                                  
Package build date            | 2020-05-23T10:49                        
Git commit ID (hash)          | 2934ac134ce141f7cc91a948cbba8e6389381fac
Git tag information           | 6.13                                    
Server supported API (Thrift) | 6.27                                    
Client API (Thrift)           | 6.27                                    
------------------------------------------------------------------------

To Reproduce

$ CodeChecker store codechecker-results/ -n firefox_$(date "+%Y%m%d%H%M") --url firefox
[INFO 2020-05-23 10:41] - Storing analysis results for run 'firefox_202005231041'
[INFO 2020-05-23 10:41] - Checking local passwords or tokens in /var/lib/jenkins/.codechecker.passwords.json
[INFO 2020-05-23 10:41] - Password file not found.
[INFO 2020-05-23 10:41] - Checking for local valid sessions.
[INFO 2020-05-23 13:27] - Storage failed: '>' not supported between instances of 'float' and 'NoneType'

Happy to help debugging

(it would be nice to show the backtrace)

[csordasmarton]

@sylvestre Could you please run your store command in debug mode (--verbose debug) and send us your output.

[sylvestre]

I did it and it took more than 13h to get this error:

[DEBUG][2020-05-25 07:52:04] {system} [15707] <140713747179328> - store.py:366 assemble_zip() - Copying file '/var/lib/jenkins/workspace/firefox-codechecker/codechecker-results/DecryptJob.cpp_clang-tidy_9ee326367450416a9885397e67a737f1.plist' to ZIP assembly dir...
[INFO][2020-05-25 08:09:14] {system} [15707] <140713747179328> - store.py:646 main() - Storage failed: Can't close the ZIP file while there is an open writing handle on it. Close the writing handle before closing the zip.
Exception ignored in: <function ZipFile.__del__ at 0x7ffa7842fea0>
Traceback (most recent call last):
  File "/usr/lib/python3.7/zipfile.py", line 1789, in __del__
    self.close()
  File "/usr/lib/python3.7/zipfile.py", line 1798, in close
    raise ValueError("Can't close the ZIP file while there is "
ValueError: Can't close the ZIP file while there is an open writing handle on it. Close the writing handle before closing the zip.

which is of course different ... :)

My codechecker-results directory is 257 Go. I am trying to run it again if I get the same error again

[gyorb]

Could the current way of extracting the zip file on the server side cause this problem? It it written out to the disk and immediately read back for extraction.
https://github.com/Ericsson/codechecker/blob/9b25c841942b8992a808bda831f8eee8eab137f1/web/server/codechecker_server/api/report_server.py#L505-L525

Probably the name should be used to open for read:

         with zipfile.ZipFile(zip_file.name, 'r', allowZip64=True) as zipf: 

Based on the documentation on Unix platforms it can be opened a second time again.

Whether the name can be used to open the file a second time, while the named temporary file is still open, varies across platforms (it can be so used on Unix; it cannot on Windows NT or later).

@sylvestre on which platform do you try run it?

[sylvestre]

I tried again in Debug and yeah, same pb again with zip.

on which platform do you try run it?

@gyorb on Debian

[sylvestre]

@gyorb did something change in "CodeChecker store" recently? It used to be pretty fast. Now, it has been running for more 10 hours (and still running)

[gyorb]

In the latest release v6.12.0 there were some clang-tidy configuration changes to show results from the header files. It can increase the found reports. Are there a lot more reports after that change? You can switch back to the old behavior with this analyzer configuration: CodeChecker analyze compile_command.json --analyzer-config clang-tidy:HeaderFilterRegex=\"\"

Could you try to run an analysis with this configuration?

[irishrover]

I faced with it. Seems codechecker/web/client/codechecker_client/cmd/store.py:324

                if sc_handler.has_source_line_comments(file_path, report_line):
                    file_hash_with_review_status.add(file_hash)

reading every single file takes a long.

[gyorb]

That part was not touched recently but maybe the increased report count causes to check a lot more files than before.

We will need to check what can be improved there.

[gyorb]

I'm working on some performance improvements for the store command.

[sylvestre]

@gyorb sorry it took forever for me to get back to this (as it takes almost a full day to run) Using

CodeChecker analyze compile_command.json --analyzer-config clang-tidy:HeaderFilterRegex=\"\"

indeed made the build completed.

I will try to run store in verbose mode but it takes several days to run :/

[gyorb]

@sylvestre there were some performance improvements done for the store did you have time to try it out?