recollindex stuck with 100% CPU usage, do not stop when interrupted

[Romano Giannetti]

I have the recollindex task "stuck" on a simple doc (that I can print easily with antiword) since 12 hours.

[romano:~/.recoll] % ps augx | grep recolli
romano 23475 95.0 0.6 107452 52040 pts/8 RN+ nov20 715:49 recollindex

The thing is stuck with:

[romano:~/.recoll] % cat idxstatus.txt
phase = 1
docsdone = 65117
filesdone = 24724
dbtotdocs = 289586
fn = /home/romano/management/asignaturas_ea/2006-07/fichas_mias/Ampliación_de_electrónica_4ºIIND_ficha.doc

If I send a CTRL-C, I have:

^CGot signal, registering stop request

...since half an hour ago. I will kill the thing -KILL now. Any hint?

Version:

[romano:~/.recoll] % apt-cache policy recoll
recoll:
  Installed: 1.17.3-2
  Candidate: 1.17.3-2
  Version table:
 *** 1.17.3-2 0
        500 http://archive.ubuntu.com/ubuntu/ trusty/universe amd64 Packages
        100 /var/lib/dpkg/status

[Alessandro Volpi (volpial3)]

recollindex has always been very slow with ubuntu. Much better on fedora.

When run with the -z option recollindex does what one would expect. It creates the data base file and terminates correctly.

When trying to update the database the program behaves as described by Romano Giannetti.

It logs a large number of errors ( example
:2:../rcldb/rcldb.cpp:1470:Db::needUpdate: error while checking existence: Error reading block 3159687168: got end of file ) and adds to the database the files which should be already registered, since they were not modified since the creation of the database.

Herein attached is a compressed file with the logs of database creation and update.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Alessandro Volpi (volpial3)]

May the bug be related to the coexistence of python-2 and python-3 on the same system ?