> Another thing that seems to slow the site down, is the number of users in
> the elog notification list (those who've subscribed). When you save a log
> entry, it takes around 30sec or longer for it to actually complete the
> save. If I remove the list of users from the notification list and just
> have a few, the save is very fast.
This problem will be fixed in version 2.2.5. Prior to 2.2.5, individual
emails were sent to all recipients. Since each email takes 0.5-1 sec., this
procedure can be very long. From 2.2.5 on, only one email is sent, but to
all recipients. The disadvantage of this method is that the "Mail to:" field
contains the email addresses of all recipients, so each recipient knows the
addresses of the other, which is maybe not always what you want. I put a new
option to discard the "Mail to:" field, but some systems the consider the
mail with a missing "Mail to:" field as spam mail. 2.2.5 will be released in
a couple of days.
> 2600 entries is too much for this application as it load the all files
> in computer memory
> expand the server memory
> Are you running on linux or Windows ? I suggest linux (faster)
> We are working on the C source to move all data from flat to database like
> SQL or mysql
> when a parameter flag like 'status' = "OK" for instance
> I suggest also to split in several logbook
> but this is depend on your 'ELOG' parametrisation and logics
>
> If your data are not 'sensitive' I can check on my linux server
>
> Etienne
It is not correct that all files are loaded into memory. Only the index
resides in memory, the data stays on disk. In my environment, I see no speed
difference between Windows and Linux. Moving to SQL will certainly not speed
up the responsiveness in my opinion. So before working on that, create a SQL
database with your 2600 entries and see how fast you can make queries on
them.
The problem with the slow response is new to me. Other users mentioned no
problem with logbooks with several throusand entries (except for the "find"
command). But I will have a look myself in the next feature and see if I can
make things better.
- Stefan |