need less verbose logging, more efficient prerequisite processing

Bug #667482 reported by siznax
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Archive Widecrawl
Confirmed
Low
siznax

Bug Description

currently, always get verbose output, eg

    MANIFEST exists: /3/incoming/newscrawl/NEWS-20100901201302653-00001-00019-ia360914/MANIFEST
    ... (repeated many times) ...
    LAUNCH file exists: /3/incoming/newscrawl/NEWS-20100901201302653-00001-00019-ia360914/LAUNCH
    ... (repeated many times) ...
    TOMBSTONE exits: /3/incoming/newscrawl/NEWS-20100914052420163-00159-00177-ia360914/TOMBSTONE
    ... (repeated many times) ...

Tags: drain
Revision history for this message
siznax (siznax) wrote :

this also effects the speed of processing as the number of "finished" uploads increases.

each pass over the prerequisites (PACKED, MANIFEST, LAUNCH/TASK, TOMBSTONE) can take a while on a busy crawler, and can make for unnecessarily large log files if you tee draintasker stdout/err to a file.

i've created a script to move aside tasks containing a TOMBSTONE file to an "_done" directory alongside the xfer_job_dir, e.g.

xfer_dir => /1/incoming/newscrawl
done => /1/incoming/newscrawl_done

this can speed processing, and make logs smaller.

we may want to include this in normal processing. if so, we'll want to also update "get-status" to look for this _done series.

Changed in archivewidecrawl:
status: New → Confirmed
assignee: nobody → siznax (siznax)
importance: Undecided → Low
summary: - need less verbose logging option
+ need less verbose logging, more efficient prerequisite processing
Revision history for this message
siznax (siznax) wrote :

oops, the script is:

eject_tombstones.sh
------------------------------------------------------------------------
r48 | steve | 2010-11-18 00:05:56 +0000 (Thu, 18 Nov 2010) | 1 line

move TOMBSTONES out of the way of drain processing

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.