MySQL Replication Configuration

Next topic

Author	Message
Ted Prosser	Sunday 06 November 2005 4:25:37 am I am trying to set up a secure eZ publish configuration with three servers and MySQL replication and rsync and I was hoping I could get some suggestions. Currently, I have two front-line web servers for serving public requests. A load balancer alternates traffic between the two. The front line servers are set up to use a local MySQL slave database for reads and a non-public administrative server (the MySQL master) for writes. The admin server is the only box with an eZ admin web site, which is not public. The front-line servers are getting constant database updates through MySQL replication. Rsync is set up to push out changes to /design, /var, and a few others, every 10 minutes or so. Cache folders are not set to synchronize. The eZ pages seem to render properly on the front-line servers, but I have the following questions: 1) Do the the front-line display-site-only servers really need to be able to write to the eZ publish database at all? I would prefer not to allow this. 2) Would the display sites ever really need to make changes to /var that would need to get copied over to the other servers? Remember that these servers are never going to run the admin site and I don't intend to use eZ publish for anything more than displaying content (no commerce). It makes sense to me that database access in this case should be read-only. I would rather avoid the eZ clustering solution for rsync if I can (I try to avoid using "beta" versions and I want to keep things as simple as possible). 3) I experimented with setting up an array of MySQL slave databases, but I was disappointed to find that eZ publish appears to randomly choose a slave database for reads. When a MySQL connection attempt fails for whatever reason, it looks the page load simply fails. eZ does not appear to be smart enough to try a different slave server on a connection failure. This seems to defeat the whole purpose of adding MySQL slaves to increase availability. Has anyone else addressed this? Thanks in advance for any replies.
Ted Prosser	Friday 11 November 2005 5:56:03 am Looks like eZ publish has to have write access to a MySQL database no matter what you are doing with the front-end. The ezsession table gets updated constantly. We've also noted temp tables get created and dropped. So much for that.

Author

Message

Sunday 06 November 2005 4:25:37 am

I am trying to set up a secure eZ publish configuration with three servers and MySQL replication and rsync and I was hoping I could get some suggestions.

Currently, I have two front-line web servers for serving public requests. A load balancer alternates traffic between the two. The front line servers are set up to use a local MySQL slave database for reads and a non-public administrative server (the MySQL master) for writes. The admin server is the only box with an eZ admin web site, which is not public.

The front-line servers are getting constant database updates through MySQL replication. Rsync is set up to push out changes to /design, /var, and a few others, every 10 minutes or so. Cache folders are not set to synchronize.

The eZ pages seem to render properly on the front-line servers, but I have the following questions:

1) Do the the front-line display-site-only servers really need to be able to write to the eZ publish database at all? I would prefer not to allow this.

2) Would the display sites ever really need to make changes to /var that would need to get copied over to the other servers? Remember that these servers are never going to run the admin site and I don't intend to use eZ publish for anything more than displaying content (no commerce). It makes sense to me that database access in this case should be read-only. I would rather avoid the eZ clustering solution for rsync if I can (I try to avoid using "beta" versions and I want to keep things as simple as possible).

3) I experimented with setting up an array of MySQL slave databases, but I was disappointed to find that eZ publish appears to randomly choose a slave database for reads. When a MySQL connection attempt fails for whatever reason, it looks the page load simply fails. eZ does not appear to be smart enough to try a different slave server on a connection failure. This seems to defeat the whole purpose of adding MySQL slaves to increase availability. Has anyone else addressed this?

Thanks in advance for any replies.

Ted Prosser

Friday 11 November 2005 5:56:03 am

Looks like eZ publish has to have write access to a MySQL database no matter what you are doing with the front-end. The ezsession table gets updated constantly. We've also noted temp tables get created and dropped. So much for that.

eZ debug

Timing:	Jan 19 2025 03:22:27
Script start
Timing:	Jan 19 2025 03:22:27
Module start 'layout'
Timing:	Jan 19 2025 03:22:27
Module start 'content'
Timing:	Jan 19 2025 03:22:29
Module end 'content'
Timing:	Jan 19 2025 03:22:29
Script end

Main resources:

Total runtime	1.6254 sec
Peak memory usage	4,096.0000 KB
Database Queries	52

Timing points:

Checkpoint	Start (sec)	Duration (sec)	Memory at start (KB)	Memory used (KB)
Script start	0.0000	0.0083	589.2734	152.6406
Module start 'layout'	0.0083	0.0062	741.9141	39.4766
Module start 'content'	0.0145	1.6095	781.3906	425.7578
Module end 'content'	1.6241	0.0013	1,207.1484	8.1484
Script end	1.6254		1,215.2969

Time accumulators:

Accumulator	Duration (sec)	Duration (%)	Count	Average (sec)
Ini load
Load cache	0.0047	0.2889	16	0.0003
Check MTime	0.0014	0.0832	16	0.0001
Mysql Total
Database connection	0.0007	0.0441	1	0.0007
Mysqli_queries	1.5801	97.2121	52	0.0304
Looping result	0.0005	0.0325	50	0.0000
Template Total	1.5654	96.3	2	0.7827
Template load	0.0019	0.1155	2	0.0009
Template processing	1.5636	96.1943	2	0.7818
Template load and register function	0.0002	0.0098	1	0.0002
states
state_id_array	0.0016	0.0978	1	0.0016
state_identifier_array	0.0044	0.2686	2	0.0022
Override
Cache load	0.0015	0.0950	13	0.0001
Sytem overhead
Fetch class attribute can translate value	0.0008	0.0505	1	0.0008
Fetch class attribute name	0.0020	0.1216	2	0.0010
XML
Image XML parsing	0.0001	0.0068	1	0.0001
class_abstraction
Instantiating content class attribute	0.0000	0.0006	2	0.0000
General
dbfile	0.0007	0.0416	10	0.0001
String conversion	0.0000	0.0008	4	0.0000
Note: percentages do not add up to 100% because some accumulators overlap

Templates used to render the page:

Usage	Requested template	Template	Template loaded
1	node/view/full.tpl	full/forum_topic.tpl	extension/sevenx/design/simple/override/templates/full/forum_topic.tpl
2	content/datatype/view/ezxmltext.tpl	<No override>	extension/community_design/design/suncana/templates/content/datatype/view/ezxmltext.tpl
2	content/datatype/view/ezxmltags/paragraph.tpl	<No override>	extension/ezwebin/design/ezwebin/templates/content/datatype/view/ezxmltags/paragraph.tpl
1	print_pagelayout.tpl	<No override>	extension/community/design/community/templates/print_pagelayout.tpl
Number of times templates used: 6 Number of unique templates used: 4

Time used to render debug report: 0.0001 secs