7.3 database conversion

[rolandgifford]

This is taking unacceptably long and I'm wondering if I have any alternatives or whether I am stuck at pre-7.3

I have 189000 images. cat.db is 15Gb thumbs.db is 199Gb

I first tried installing 7.3 mid August when I came back from a trip before importing the images from that trip. The db conversion ran for 40 hours before I killed it. The cat.db had about doubled (I'm not concerned about that) and the progress bar was showing about 2/3 through. Whenever I looked at Task Manager during this period I was seeing low levels of CPU and Disk activity, typically less than 5%. I killed the process because I had about 20,000 images from my trip to import and needed to get started.

I have now almost finished importing/cataloguing/deleting from those and have a gap in my workflow where I don't need the PS catalogue so decided to start the conversion again. After 48 hours I killed it again as it was showing less than 50% on the progress bar. Machine activity about the same as before, cat.db more or less doubled again which happened during the first 12 hours.

I have no way of telling how much longer the process would take so am reluctant to start again as I may still be waiting for it to complete after 5 days. Unfortunately the PS changes requiring this change are of no interest to me. I search on labels and sometimes the top level statuses (e.g. Not GEO Tagged) and don't search on the details section at all. I do have the concern as well that 7.3 will be even slower importing. I stopped versioning and labeling some time ago as the stand alone process for both is far faster than doing it when importing.

Is there any alternative to the db conversion which would be quicker? I make extensive use of versioning, much of which doesn't fall within rules as I don't always use the same name format.

[Hert]

The database upgrade is main that: database tasks. Cpu usage will be low during the upgrade because the computer mainly waits on the disk.
I don’t understand why you kill it after 48 hours while there’s still processor activity. Btw. At 50% after 48 hours does not mean it takes another 48 hours. The progress gives the steps needed to complete…some steps need more time to complete than others.
Just don’t kill the upgrade.

This is the biggest database upgrade ever and unfortunately takes time…depending on the hardware a long time.
Fwiw; I have catalogs here that are a multitude in volume compared to your catalog and all completed the upgrade within a few hours… the time needed mainly depends on the hardware.

Make sure that your database is on your fastest disk, an internal SSD works best. And make sure that your catalog folder is excluded from any virus scanning software, including virus scanners that come integrated with the OS (like Windows Defender). Virus scanners can slow down a SQLite database significantly due to the fact that SQLite is constantly creating journal files which virus scanners should not touch.

Summarized:
1. Use your fastest interval SSD disk available to store your catalog on
2. Exclude the catalog folder from virus scanners
3. Give it time

The only alternative (assuming you keep files in-sync) is to start a new catalog, enable ICS reading, import everything to the new catalog, disable ICS reading. The pro for this approach is that you can do it gradually over time.

[rolandgifford]

I killed it after 48 hours with the progress bar showing less than 50% complete as that implies more than 48 hours to go. It may be inaccurate and actually means 30 hours (but could mean 60 hours) but is an accurate indication of the time left. A progress bar which shows steps remaining, some of which are short, is of little use. I assume yours has been written to be useful. I have tasks including PS cataloguing completion which I have to finish by the end of Tuesday this week. I don't have lots of time to waste.

I will use your suggestions where possible when I next have the opportunity for significant downtime, probably late October.

Does the ICS approach not lose versioning data. I was aware of it and had discounted it on that assumption. I mentioned that I make use of versioning so that you would know that when suggesting options.

[Hert]

Versions are part of ICS

[rolandgifford]

Thanks, I'll consider that.

I'll see if I can allocate an SSD drive with enough space as the sensible first option.

How much space is needed? I remember space issues in the past for database compacting and having to set the temp folder to somewhere with enough space, I assume that the same applies with this

[Hert]

1. Copy the cat.db file only (not the thumbs.db) to a temp folder on your SSD
2. Start psu and in preferences, enable the login dialog
Close PSU
3. Start psu and in the login dialog select the copied file as the database
4. Complete the upgrade
5. Close Psu
6. Copy back the cat.db file to the original folder (backup the old one)
7. Start psu

Do I read between the lines that you don’t have the catalog files on an SSD? An average SSD is about 20x faster than a HDD. Since PSU is all about database, that means that Psu will perform about 20x faster if you keep the catalog files on the SSD. That is VERY significant. It is highly recommended to keep the catalog files on an internal SSD. And exclude the catalog folder from virus scanners

[rolandgifford]

You are correct that I don't keep my catalogue files on an SSD. The PC that I use for PS doesn't have an SSD fitted but I have other PCs that do.

THE PC I use for PS is my old main PC so it will get an SSD the next time that I upgrade my main PC. Windows tells me that I need to do that for Windows 11 but I have been ignoring it. I generally upgrade everything at the same time rather than adding individual upgrades but have noted your SSD comments and will consider.