steven l. rubenstein wrote:
Half the time I click on "Save Page" the next screen is a "Preview" (and my edits are not yet saved). I have been careful about this, and am definitely clicking on "Save Page," not "Show Preview." And this is happening a lot. Is this happening to anyone else, or is it just me? Does anyone have any idea why it is happening?
Been happening to me too, on meta as well as en. (I've been adding to [[m:Bash]].)
It's a known symptom of a very busy database - rather than committing the change, it dumps you back to the preview screen.
- d.
David Gerard wrote:
steven l. rubenstein wrote:
Half the time I click on "Save Page" the next screen is a "Preview" (and my edits are not yet saved). I have been careful about this, and am definitely clicking on "Save Page," not "Show Preview." And this is happening a lot. Is this happening to anyone else, or is it just me? Does anyone have any idea why it is happening?
Been happening to me too, on meta as well as en. (I've been adding to [[m:Bash]].)
It's a known symptom of a very busy database - rather than committing the change, it dumps you back to the preview screen.
News to me. If the database is too busy, you should get a timeout screen or a "cannot contact database server" error message.
We don't know exactly why this problem happens, but it's probably caused by loss or temporary unavailability of session data. Brion says we still got these reports when we put sessions in NFS, so memcached can't explain the problem in its entirety, but nonetheless I'd be inclined to think the current increase in reports is due to load on the memcached servers. We should probably try installing memcached on another 10 or 15 servers and see if that fixes the problem.
A closely related problem is being spontaneously logged out if you didn't select the "remember me" checkbox when you logged in.
-- Tim Starling
News to me. If the database is too busy, you should get a timeout screen or a "cannot contact database server" error message.
We don't know exactly why this problem happens, but it's probably caused by loss or temporary unavailability of session data. Brion says we still got these reports when we put sessions in NFS, so memcached can't explain the problem in its entirety, but nonetheless I'd be inclined to think the current increase in reports is due to load on the memcached servers. We should probably try installing memcached on another 10 or 15 servers and see if that fixes the problem.
A closely related problem is being spontaneously logged out if you didn't select the "remember me" checkbox when you logged in.
-- Tim Starling
I've noticed that problem since the upgrade - sometimes a second or third submit does it, sometimes it never works and I just copy the edit to notepad, close Firefox, and try again. I'd say it happens maybe every third edit. This is very different from the timeout messages.
Ian (Guettarda)
On 26/10/05, Guettarda guettarda@gmail.com wrote:
We don't know exactly why this problem happens, but it's probably caused by loss or temporary unavailability of session data. Brion says we still got these reports when we put sessions in NFS, so memcached can't explain the problem in its entirety, but nonetheless I'd be inclined to think the current increase in reports is due to load on the memcached servers. We should probably try installing memcached on another 10 or 15 servers and see if that fixes the problem.
A closely related problem is being spontaneously logged out if you didn't select the "remember me" checkbox when you logged in.
I've noticed that problem since the upgrade - sometimes a second or third submit does it, sometimes it never works and I just copy the edit to notepad, close Firefox, and try again. I'd say it happens maybe every third edit. This is very different from the timeout messages.
It was in existence before the upgrade - I had a spate of it twice in eight months or so - but was quite obscure, and I believe most people just thought "bugger, I hit 'preview' not 'save'". Anecdotally, it did start happening a lot more just after the upgrade; I don't know if that suggests a software issue or simply a load one.
-- - Andrew Gray andrew.gray@dunelm.org.uk
Tim Starling wrote:
Half the time I click on "Save Page" the next screen is a "Preview" (and my edits are not yet saved). I have been careful about this, and am
Me to - now seems fine, but last night I had the same problems.
We don't know exactly why this problem happens, but it's probably caused by loss or temporary unavailability of session data. Brion says we still got
I suspect the same, as when the above effect appears I also have problems to use the rollback function, it then gives an error message saying something about session data if I recall it correctly.
A closely related problem is being spontaneously logged out if you didn't select the "remember me" checkbox when you logged in.
Strangely, this only happened in the Thai wikipedia yesterday, while in the english one the login was remembered at the same time.
[[w:en:User:Ahoerstemeier]]
Andreas Hoerstemeier wrote:
A closely related problem is being spontaneously logged out if you didn't select the "remember me" checkbox when you logged in.
Strangely, this only happened in the Thai wikipedia yesterday, while in the english one the login was remembered at the same time.
It's not that strange, assuming it is a memcached server problem, the frequency of the problem would depend on which server your session ID is on, and you have a different session ID for th and en. That's why logging out and logging back in helps: you get a different session ID, hopefully on a different server.
-- Tim Starling