Discussion:
[ic] IC + Apache 2.4
René
2018-07-06 23:05:38 UTC
Permalink
Hi,

What is the proper way to run IC nowadays with Apache 2.4? If i got it
right, mod_interchange and mod_perl2 was intended for older versions of
apache.. Is it still vlink, or is there an other way that i have missed?

René
Jon Jensen
2018-07-06 23:47:44 UTC
Permalink
Post by René
What is the proper way to run IC nowadays with Apache 2.4? If i got it
right, mod_interchange and mod_perl2 was intended for older versions of
apache.. Is it still vlink, or is there an other way that i have missed?
Hi, René.

For most of the sites End Point works with we're using Apache 2.4 running
the compiled vlink CGI. Sometimes we also have nginx in front, serving
static files directly and proxying through to Apache for Interchange and
other CGIs.

Jon
--
Jon Jensen
End Point Corporation
https://www.endpoint.com/
René
2018-07-07 22:43:52 UTC
Permalink
Hi, René.
For most of the sites End Point works with we're using Apache 2.4
running the compiled vlink CGI. Sometimes we also have nginx in front,
serving static files directly and proxying through to Apache for
Interchange and other CGIs.
Hi Jon,

Are you always running one ic-instance per server, or do you also have
cases where you run two different versions of ic?

I have not yet had the chance to get my fresh perbrew-perl + ic up and
running, but the idea would be to try to test and trace down why i get
the strange lockups that freezes all ic-pages in all catalogs for a
couple of minutes after i update a table that runs on gdbm..

Is there any point to move the locale-db to postgres, or should i try to
stick to gdbm? If i understand the idea right, gdbm-tables are loaded
into memory, and therefore there is no db-access for each [L] / [LOC]
accurance.

Is there btw an option to use a postgres table, but on a ic-reload, the
table would be read into memory?

It has to be something messed up with my gdbm, or maybe because i run
Debian in 32bit?


René
Jon Jensen
2018-07-07 23:41:37 UTC
Permalink
Post by René
Post by Jon Jensen
For most of the sites End Point works with we're using Apache 2.4
running the compiled vlink CGI. Sometimes we also have nginx in front,
serving static files directly and proxying through to Apache for
Interchange and other CGIs.
Are you always running one ic-instance per server, or do you also have
cases where you run two different versions of ic?
We have cases where we run two different versions of Interchange on the
same server, yes.
Post by René
I have not yet had the chance to get my fresh perbrew-perl + ic up and
running, but the idea would be to try to test and trace down why i get
the strange lockups that freezes all ic-pages in all catalogs for a
couple of minutes after i update a table that runs on gdbm..
Is there any point to move the locale-db to postgres, or should i try to
stick to gdbm? If i understand the idea right, gdbm-tables are loaded
into memory, and therefore there is no db-access for each [L] / [LOC]
accurance.
I do not believe that GDBM tables are loaded into memory by Interchange,
but they probably are by the operating system via mmap or at least disk
caching.

If your site gets relatively low traffic, you'd might as well try loading
the locale table into Postgres and see how it goes.

Jon
--
Jon Jensen
End Point Corporation
https://www.endpoint.com/
Loading...