Jump to: navigation, search
MithiWiki Home > ConnectXf Home > ConnectXf Administration > Configuration > Mcsdata partition is full


Troubleshooting Icon.png
Troubleshooting
Product ConnectXf
Version 3.10
Applies to Administrators
Level Advanced




mcsdata partition is full

Symptoms

  • The LDAP and Postgres services are stopped automatically
  • df -mh shows /mcsdata partition as full
  • Users cannot login to their accounts
  • Mail flow has stopped (queues growing)

Possible causes

Related to Backups

The folder for holding backups, keeps filling up with old backup files

Related to Mail Delivery reports

Mail Delivery reports grow too large

Related to database

Database grows too large

Related to Mail archiva volume

  • In Mailarchiva application, when a current ACTIVE volume becomes full and there are no more UNUSED volumes available, then emails are no longer archived by the application. Instead, the mails are kept in no archive queue.
  • This no archive queue is often stored in following location in /mcsdata partition. As the size of this folder grows it consumes /mcsdata and eventually leads to disk full situation.

Diagnosis & Resolution

Related to Backups

Clean old backups from the backup folder

  • Confirm the the agent to flush_old_backups has the following attributes
    • on
    • Checking the contents of the right folder.
    • Keeping local backs for less than 7 days
/mithi/mcs/bin/listagents.sh | grep flush 
  • The agent should be "on"
  • The agent should be checking the patch where the backups are being taken. By default it is the /mithi/mcs/backups folder.This is the first parameter to the agent-delete_old_files.sh
  • The number of days for which the backup has to be kept is the second parameter to the agent_delete_old_files.sh script.
  • If either of these is not true, then modify the agent as follows:
/mithi/mcs/bin/modifyagent.sh flush_old_backups -status on -restartagents t -agent_params '/mithi/mcs/backups 7'
  • Confirm that the agent is being called in the log.
cat /var/log/mithi/mcs/agent_delete_old_files.sh.log | grep "backups"
  • Output should show success
  • Check the size of the other folders such as
    • /mithi/temp
    • /mithi/mcs/temp
    • /home/mcs/installables
    • /root/backups

Related to Mail Delivery reports

Normally the server generates and retains mail delivery reports for 10 days. In certain situations when the mail server has been mis-configured and mail loops with lots of duplication, or there is excessive mail traffic on a few days, the reports can grow very large. These will occupy all free space on the / or /mcsdata partition and cause the server to stop all essential services.

To confirm how much space is occupied by the mail delivery reports

du -sh /mithi/mcs/modules/mithi-bl/report/MDR/

If this is occupying a maximum part of the disk space than you should clean these

rm -rf /mithi/mcs/modules/mithi-bl/report/MDR/domains/*
rm -rf /mithi/mcs/modules/mithi-bl/report/MDR/server/*

Restart syslog, smtp, ldap and postgres services

/etc/init.d/postgresql restart
/etc/init.d/ldap restart
/etc/init.d/qmail-smtp restart
/etc/init.d/syslog-ng restart

Reduce the retention period to 5 days

/mithi/mcs/modules/mithi-bl/conf/domains/<domain>/domain_parameters.properties
Change the value of r_szMDRArciveLimit to 5
r_szMDRArciveLimit=5
Save and close the file.

Related to Database

In Connect Xf the database is used to store the data required to generate the spam reports. This can sometimes cause the database size to grow large filling up the /mcsdata partition.

To confirm whether the space occupied by the database is large, use the following command:

du -sh /var/lib/pgsql/data

Solution

  • The immediate solution is to drop the tables used for the spam reports.
  • However to do so the database service has to be started, which will be stopped. So first we must reduce the space in the /mcsdata partition. There are two ways to do this

a) Clear some other data, start pgsql and then drop the tables

b) Force start the pgsql and drop the tables.

  • The steps to force start are given in the section on the Database.
  • The report table can be dropped using the following commands:
psql REPORT
drop table mailevents;
\q
  • If the spam reports are required then empty tables need to be recreated using the following command
psql REPORT
               CREATE TABLE mailevents
               (
                       maileventid serial primary key,
                       datetime timestamp not null default now(),
                       whoexecuted varchar(64),
                       clientip varchar(15),
                       starttime varchar(64),
                       endtime varchar(64),
                       processingtime varchar(64),
                       id varchar(64),
                       command varchar(32),
                       result varchar(32),
                       outputmessage varchar(10240),
                       errormessage varchar(10240),
                       eventparameters varchar(10240),
                       status varchar(32),
                       sender varchar(128),
                       recipient varchar(64),
                       date varchar(64) default now(),
                       messageid varchar(128),
                       subject varchar(256),
                       sizeinkb varchar(8),
                       attachmentcount varchar(8),
                       attachmentlist varchar(256),
                       returncode varchar(8),
                       spam_mail varchar(8),
                       folder varchar(128),
                       processedsteps varchar(3072),
                       currentdate date not null default current_date
               );
  • Create index required by the table
create index idx_mailevents_recipient on mailevents (recipient);
\q

Related to Mail Archiva Volume

Follow the link for details on How to handle Archive Emails when the archive volumes are full.