Upgrade Storagesync takes a long time due to unnecessary pdf access
I am using zotero 20b7.6 with firefox 3.5.6 on WinXp, attachments are synched to my own WebDAV on Apache 2.2, same machine.
Right after initiating the sync, many of the pdf files in my firefox-profile zotero storage folder are accessed (didn't check whether all were accessed, but sure looks like it). It looks like each file is read beginning to end (filemon shows increasing read offset at 4K increments)
I intiate the sync just after having performed a sync, so there should actually be nothing left to synch.
The process starts out with a couple actions on zotero.sqlite-journal, but that is over quite quickly.
I have pdf indexing tools installed, but when rebuilding the index, pdftotext appears as accessing process, in case of the synch it is firefox, so that is probably unrelated.
Seems that could be something worth fixing, but a quick workaround would be appreciated as well.
Let me know if I can provide more info.
Thanks for the good work
Björn
Right after initiating the sync, many of the pdf files in my firefox-profile zotero storage folder are accessed (didn't check whether all were accessed, but sure looks like it). It looks like each file is read beginning to end (filemon shows increasing read offset at 4K increments)
I intiate the sync just after having performed a sync, so there should actually be nothing left to synch.
The process starts out with a couple actions on zotero.sqlite-journal, but that is over quite quickly.
I have pdf indexing tools installed, but when rebuilding the index, pdftotext appears as accessing process, in case of the synch it is firefox, so that is probably unrelated.
Seems that could be something worth fixing, but a quick workaround would be appreciated as well.
Let me know if I can provide more info.
Thanks for the good work
Björn
Zotero checks the mod times of all attachment files on every sync to see if files have been edited externally. As far as we've seen, this should be basically instantaneous. If the mod time doesn't match the stored version, Zotero computes a hash of the file contents, which could take longer. It's possible you're experiencing a timestamp-related bug that's causing it to check all file hashes unnecessarily. If you provide a Debug ID for the sync, we might be able to tell you more.
I performed the following steps without ever changing an item, even since yesterday.
Synching a freshly synched library takes about 1 minute.
The Debug ID is D1287051464.
Then I synched from a different computer that has the absolute same library.
That took about 5 minutes
Then Synched the first computer for about 1 minute
The Debug ID is D1083847754.
But after that the synch takes only seconds?!
The Debug ID is D2117623446.
Then I synched from the other again.
Then again from the first computer, still seconds!
The Debug ID is D2105256969.
I fixed it :-)
If it stays like that I am fine. Still sending you all those IDs, maybe it helps.