Low performance throughout OMERO (and some other questions)

[morgan]

Hi,

I'm new to OMERO and just installed the server on my personal computer (2,8Ghz i7, 8 GB RAM) running Mac OS X 10.8.2 using homebrew. I'm using ICE 3.4.2 and everything seems to be working. Except, performance is horrible.

# Importing a folder of 20 TIF images ~2MB each took 20-30 minutes. Times were even worse for ND2 images, but I guess that's understandable since support for those isn't 100% if I'm not mistaken.
# Opening the image viewer takes 2-5 minutes. Clicking buttons in image viewer and waiting for response takes 10-30 seconds.
# Everything is unusably sluggish and this is with relatively small images - no stacks or anything. I used Java 6 first and tried updating to Java 7 without noticing any difference in speed.

Do you have any ideas of what could be causing this?

Another question I have is if it's possible to assign colors to channels, rename channels and such. The main reason for this is that the ND2 images gets imported with the wrong colors since the order of the channels in the ND2 files isn't RGB. Also, for 4-channel images two channels gets the same color (blue), whereas I'd like it to be for example yellow or cyan.

I've read in some other post that the importer looks at the metadata and tries to determine the channels according to emission wavelength. The property in the ND2 files however is named "CH# {Laser Wavelength} #1" and it's value e.g. "488 {LaserPower}: 1.0". Is it possible to make the importer read these properties properly, or would this be a more BioFormats-related question?

[morgan]

if it's possible to assign colors to channels

So, I managed to find out that this is possible by right-clicking on the channel in the image viewer. I'd therefore instead like to ask if it's possible to batch-change channels like this (since it otherwise would be extremely time consuming). This of course would not be needed if it was possible to correct on import, however probably still useful. The question about renaming channels remains and batch-renaming would very much be preferred.

Thanks in advance!

[wmoore]

To answer your channel colour & naming issues, you can now (4.4.6 release) rename channels, and there is an 'apply to all' button in the right Preview panel for applying rendering settings to all images in a dataset. See http://www.openmicroscopy.org/site/supp ... -data.html

You can also right-click on a thumbnail and "Copy Settings" then right click on the same or different dataset in the tree on the left and "Paste Settings".

BioFormats (used in import) does attempt to read the Channel colours from some formats. If you want us to try and read the data from your files, you could upload an example file at http://qa.openmicroscopy.org.uk/qa/upload/ then reply here with the upload ID and the channel colours you're expecting - then we'll see what we can do.

Cheers,

Will.

[kennethgillen]

And to address your performance question:

I'd like to ask for server and client logs, and an upload of examples of your data if possible.

Log server files will be:
/home/omero/OMERO.server/var/log

The Insight client will be logging to:
$HOME/omero/log

If you could please send us these and an example of each of the file types via the following facility:
http://qa.openmicroscopy.org.uk/qa/upload

That way we can see if there is anything obviously going wrong from the server logs, and we can see what our import times are like against our own OMERO installations.

Many thanks,

Kenny

[morgan]

Log files:
https://mega.co.nz/#!J0hn2SIT!Bo4rl1uU_ ... o8A1yOsukc
https://mega.co.nz/#!AxxgSQhB!Ep-w5xSpk ... MFa6gthqcc

I've uploaded a ND2 file with channels for DAPI, 488, 568 and 647.
http://qa.openmicroscopy.org.uk/qa/feedback/4969/

Usually, opening this with BioFormats in ImageJ has switched the blue and red channel. I don't know if it was recently updated, or if something is different on my home computer compared to my portable (which I won't have access to until monday), but opening the picture today assigned the colors properly (in ImageJ).

Opening the image in Omero's image viewer (that takes about 5 to 7 minutes btw), the channels are named according to the metadata but DAPI is red and 568 is blue. The far red channel is blue as well, and I'd love for it to be any other color (white, yellow, cyan - doesn't really matter, though I'd prefer if I could set a default somewhere).

Import seems to be extra slow with ND2 images, and I haven't had the same problem with .dv stacks from DeltaVision or .lif "experiments" from Zeiss software. Handling the images is extremely slow for all image types though.

[wmoore]

Thanks for the file and Channel Color info. Ticket created here: http://trac.openmicroscopy.org.uk/ome/ticket/10530

[jburel]

Hi Morgan

Few users have reported issues with .nd2 files and some problems have already be fixed
see https://github.com/openmicroscopy/bioformats/pull/420
Those changes will be available in the next release of bio-formats.

Regards

Jmarie

[irradiator]

Hi, I am curious to know, whether the problem related to the low performance was solved.

I have installed OMERO (using homebrew) on a machine running OS X 10.7 (in early May 2013) without any problems and later on 2 machines running OS X 10.8 where if first had problems to run the server (I was always getting the error message:
"could not connect to server: No such file or directory
Is the server running locally and accepting
connections on Unix domain socket "/var/pgsql_socket/.s.PGSQL.5432"?"

After solving the problem using http://jaygoldman.com/2012/11/fixing-postgres-connection-errors-on-mountain-lion/ I could start the server and everything seems to work fine - but with an unreasonable low performance (as discribed below).

Does anybody have a clue how I can fix the problem?
Thanks in advance from Canada,
Hilmar.

[kennethgillen]

Dear Hilmar,

Best regards from Scotland!

Are you having issues with a particular image type, or all image types?

I am running a local vanilla OSX OMERO.server which was also installed via Homebrew, following the OMERO Mac OS X installation walk-through with Homebrew on the openmicroscopy website, and it's performance is pretty good, so there must be something going awry in the configuration.

As usual for all performance matters, I'd like to ask for a copy of the server and client logs please, and a copy of the diagnostics (which will also include OMERO version).

Server log files will be in a subdirectory of the folder where the OMERO server binaries are deployed, e.g.
/usr/local/Cellar/omero/4.4.8/var/log/ (depending of course on your OMERO.server version)

The Insight client will be logging to the home directory of the user running the client, e.g.
$HOME/omero/log

And can I also please have a copy of the output (sanitised for confidential config settings) of the following command:

Code: Select all

omero admin diagnostics

Many thanks,

Kenny

[irradiator]

Hi Kenny,
thanks a lot for taking your time to have a look at my log files.
I have attached an archive containing all requested files.

However I have messed up my installation on my MacBook (the computer I am using right now) so much (probably only the environment variables) that the omero server doesn't start any more (Python quit unexpectedly while using the Python plug-in). I think I can fix that once I am back from my trip to Vancouver (where I am sitting now). Since I installed omero on 2 computers running Mountain lion that show the same bad performance I have extracted the diagnostics file from my MacMini. In principle the installation on the MacBook and the MacMini should suffer one and the same problem.

As I mentioned in my last post I had to use the information that were posted here http://jaygoldman.com/2012/11/fixing-po ... tain-lion/
in order to run the server. Since I am not that proficient in using UNIX and controlling my computers from the shell that could probably be something that might have been not too good.

Thanks a lot again, Kenny for having a look at the files. Greetings from Vancouver, BC,
Hilmar.