[Bug 1044] pmchart very slow when Overview-panning 300MB 14-day archive file

[bugzilla-daemon@xxxxxxxxxxx]

     Ken McDonell
 changed
              bug 1044

What	Removed	Added
CC		kenj@internode.on.net

Comment # 2 on bug 1044 from Ken McDonell

Frank,

Are the long forwards and backwards reads for a specific metric, or for more
than one metric?

Unfortunately without the archive these sort of problems have proven
nigh-on-impossible to debug in the past ... in this case I'd need the good and
the bad archive.

Looks like reading to a mark record or reading to end of archive to me, which
sometimes is necessary to find a bound (that may not exist in the archive) for
a COUNTER or INSTANT metric.  But the code is supposed to remember the state of
these scans, so doing it multiple times is definitely broken.  We have lots of
QA coverage in this area, so this is probably a subtle corner case.

---

      You are receiving this mail because:

• You are on the CC list for the bug.