Scenario 21: Delete SPW

[jburel]

• Delete Plate
• Delete Plate Acquisition
• Delete Screen

---

insight: possible to delete plate, plate acquisition, and screen

---

web:

• plate acquisition not visible so not possible to delete.
• Cannot delete plate Internal error

---

server:

• primary issue is very slow deletes on nightshade. (see below for info)
• possible solutions:
  • upgrade to 8.2.16 on nightshade
  • try rebuilding/recreating indexes on nightshade

[jburel]

Referencing ticket #3048 has changed sprint.

[jburel]

Moved from sprint 2010-10-28 (18)

[cxallan]

OMERO.web: OMERO.web 4.2.1-DEV-r8430-trunk1169 tested fine apart from a similar display issue to #3092.

[cxallan]

Details from delete performance testing:

lrwxrwxrwx  1 omero omero        38 2010-11-02 17:10 OMERO-CURRENT -> OMERO.server-4.2.1-DEV-r8481-trunk1190

omero@gretzky:~$ time OMERO-CURRENT/bin/omero -s localhost -u cxallan delete Plate:451
Using session 4ceb8e9b-f362-4eeb-8b9f-edf21e4f9676 (cxallan@localhost). Idle timeout: 10.0 min. Current group: default
Deleting Plate (451L,)... ok.

real	4m40.640s
user	0m0.350s
sys	0m0.120s

2010-11-03 12:01:35,327 INFO  [org.perf4j.TimingLogger] (l.Server-4) start[1288785415497] time[279830] tag[omero.delete.tx]


lrwxrwxrwx  1 jboss  jboss         38 Nov  2 22:11 OMERO-CURRENT -> OMERO.server-4.2.1-DEV-r8486-trunk1191

jboss@nightshade ~ $ time OMERO-CURRENT/bin/omero -s localhost -u cxallan delete Plate:54 
Password:
Created session 67dfe46b-4b60-46ae-aaa4-d26687fc0734 (cxallan@localhost). Idle timeout: 10.0 min. Current group: default
Deleting Plate (54L,)... ok.
Traceback (most recent call last):
  File "/usr/lib/python2.4/logging/__init__.py", line 737, in emit
    self.stream.write(fs % msg)
ValueError: I/O operation on closed file

real	26m51.814s
user	0m0.452s
sys	0m0.048s

2010-11-03 12:38:49,173 INFO  [org.perf4j.TimingLogger] (.Server-12) start[1288786328156] time[1601017] tag[omero.delete.tx]

[cxallan]

Both tests done using gretzky:~jburel/16x24x3Plate.ome (16x24; 384-well; 3 field plate)

[jmoore]

Major difference I see in diffing the two schemas is:

CREATE INDEX channel_logicalchannel ON channel USING btree (logicalchannel);
CREATE INDEX channel_pixels ON channel USING btree (pixels);
CREATE INDEX channel_statsinfo ON channel USING btree (statsinfo);

i.e. nightshade is missing these 3 indexes.

Nightshade has 255K logical channels, while gretzky has 33K.

A couple of things to try would be:

• remove the indexes from gretzky and compare
• add the indexes to nightshade and compare
• load gretzky with a similar number of logical channels and compare

[jmoore]

Performance Statistics   02:06:40 - 16:00:00
Tag                                                  Avg(ms)         Min         Max     Std Dev       Count
...
omero.delete.Channel                                     0.6           0          56         1.1        3456
omero.delete.ChannelBinding                              0.4           0          12         0.5        3456
omero.delete.DetectorSettings                          102.3           0         306       144.0           3
omero.delete.Image                                      33.0          32          44         0.8        1152
omero.delete.ImagingEnvironment                         69.1          41         122        23.5        1152
omero.delete.LightPath                                 145.0          64         304        70.1        3456
omero.delete.LightPathEmissionFilterLink                 0.5           0          18         0.6        3456
omero.delete.LightPathExcitationFilterLink               0.4           0          10         0.6        3456
omero.delete.LightSettings                             300.3         300         301         0.5           3
omero.delete.LogicalChannel                              0.6           0          37         0.8        3456
omero.delete.Pixels                                    195.5         113         353        70.9        1152
omero.delete.PlaneInfo                                   0.6           0          11         0.5        3456
omero.delete.Plate                                      56.0          56          56         0.0           1
omero.delete.QuantumDef                                 50.5          23         103        23.7        1152
omero.delete.RenderingDef                                0.8           0           8         0.5        1152
omero.delete.StageLabel                                 70.3          42         123        23.7        1152
omero.delete.StatsInfo                                   0.6           0           3         0.5        3456
omero.delete.Thumbnail                                   0.6           0           2         0.5        1152
omero.delete.Well                                        0.7           0          40         2.1         384
omero.delete.WellSample                                  0.4           0          17         0.8        1152

[jmoore]

The indexes were probably a red-herring.

Other factors that may be involved:

• nightshade 8.2.4 v. gretzky 8.4.4
• vaccuuming?

[cxallan]

Password:
Created session bca471a0-a230-4d0b-9df9-02ed0c374ce5 (cxallan@localhost). Idle timeout: 10.0 min. Current group: default
Deleting Plate (102L,)... ok.

real 6m24.265s
user 0m0.370s
sys 0m0.190s

Nightshade database, same Plate.

[cxallan]

Woops.

omero@howe:~$ time OMERO-CURRENT/bin/omero -s localhost -u cxallan delete Plate:102
Password:
Created session bca471a0-a230-4d0b-9df9-02ed0c374ce5 (cxallan@localhost). Idle timeout: 10.0 min. Current group: default
Deleting Plate (102L,)... ok.

real	6m24.265s
user	0m0.370s
sys	0m0.190s

Nightshade database, same Plate.

[jburel]

Moved from sprint 2010-11-11 (19)