Editing Category:Troubleshooting/Punjabgovt Project

[[category:Troubleshooting]]

'''Zimbra Server 8'''

Architecture :

1. MTA : 192.168.70.54
 
                           Cluster IP : 192.168.70.53

2. MTA : 192.168.70.55

Mailbox server (Failover)

1. Mailbox : 192.168.70.56

2. Mailbox : 192.168.70.57

Archive Server : 192.168.70.60   (virtual machine)

That is situated in 192.168.70.55 

1. Zimbra server cluster Ip is not working but zimbra serices are running:

Command : 

1. clustat status

2. On both the server's simultaneously :

service rgmanager stop

service cman stop

then

service cman start

service rgmanager start

3. Where Archive server located , at 192.168.70.55 that is MTA2




== == '''Punjab Government Servers - Startup and shut-off Commands''' == ==





'''SAMBA SERVER SHUTDOWN PROCESS'''


This server running Samba PDC,DNS,DHCP,LAM,ntp.

SERVER MASTER ADDRESS : 192.168.10.81

SERVER SLAVE ADDRESS : 192.168.10.82

//Run these command on both master and slave simultaneously as root

# service ctdb stop 

# service clvmd stop

# service cman stop


'''SAMBA SERVER STARTUP  PROCESS'''

#service cman start

#service clvmd start

<check partitions are mounted or not ,If not mounted then (mount –a)

#mount -a

# service ctdb start

For testing run

# ping 192.168.10.93 //cluster ip

# ctdb status 

'''
ZIMBRA MAIL SERVER SHUTDOWN PROCESS'''


There are five servers of Zimbra : 

1.192.168.70.54 – MailBox1

2.192.168.70.55 – MailBox2

3.192.168.70.56 – MTA1

4.192.168.70.57 – MTA2

5.192.168.70.60 – ZIMBRA ARCHIVING

Cluster IP : 192.168.70.53(mail.punjabgovt.gov.in)

'''STEPS:'''

1.Firstly shutdown properly 192.168.70.60 that is virtual machine inside the 192.168.70.55

Command run as a root user

# init 0

2.Then shutdown MTA servers properly 192.168.70.56 and 192.168.70.57

Command run as a root user

# init 0

3.Then properly shutdown MAILBOX servers 192.168.70.54 and 192.168.70.55

Command run as a root user

# init 0

'''
 ZIMBRA MAIL SERVER STARTUP PROCESS'''

1.Start automatically both MTA servers

2.Mail Box 2 server : 192.168.10.55

Run command as a root user,

Command : ifup bridge0

3.Then on both the machine(mailbox1 and mailbox2 run simultaneously )

Command run as a root user,

#service cman start

#service rgmanager start

4.Check if cluster is in mailbox1

# clustat

5.If cluster is not in mailbox1

Then run on machine (mailbox1 )

# clusvcadm –d mailCluster

#clusvcadm –e mailCluster –m `hostname`



== 
'''DOMAIN SERVER SHUTDOWN PROCESS''' ==

There are two domain servers that contain 4 virtual machines each so firstly properly shut down these virtual machines on both the servers then after shutdown the base machines of domain1 and domain2

Virtual Machines Domain1 

Domain1 : 192.168.10.57

Virtual Machine 1 :

1.AAA 1: 192.168.10.92

2.SQUID1: 192.168.10.94

3.LDAP1: 192.168.10.96

4.NITROBIT1: 192.168.10.95

Domain2: 192.168.10.58

Virtual Machine 2 :

1.AAA 2 : 192.168.10.97

2.SQUID2 : 192.168.10.99

3.LDAP2 : 192.168.10.96

4.NITROBIT2 : 192.168.10.100

CLUSTERS IP:

1.LDAP : 192.168.10.98

2.SQUID : 192.168.10.105

3.AAA : 192.168.10.104

4.NITROBIT ARE IN FAIL OVER ENVIRONMENT

How to Shutdown these servers properly

'''Steps:'''

1.Properly shutdown ldap,squid,aaa  (run below commands simultaneously) 

Command run as a root user

# service rgmanager stop

# service cman stop

# init 0 

2.On domain 1 and domain 2 machine (run simultaneously on both the machines)

Run command as a root user

#service rgmanager stop

#service cman stop

'''DOMAIN SERVER STARTUP PROCESS'''

1.Firstly on Domain1 machine then after 2-3 minutes Domain 2 machine.

2.On Domain 2 machine run

# ifup bridge0

3.Then on both Domain1 and Domain2 machines run command simultaneously)

#service cman start

#service rgmanager start

4.Now all the base machines are up

5.Then Check virtual machine (ldap,aaa,squid,nitrobit)

6.On LDAP1 machine run command to check cluster is in LDAP1 or LDAP2.

# clustat

If machine is in LDAP2 then run command on LDAP1 machine

# clusvcadm –d dummy_httpd

#clusvcadm –e dummy_httpd –m `hostname`

Check by ping 192.168.10.98

7.On SQUID1 machine run command to check cluster is in SQUID1 or SQUID2.

# clustat

If machine is in SQUID2 then run command on SQUID1 machine

# clusvcadm –d Squid

#clusvcadm –e Squid –m `hostname`

Check by ping 192.168.10.105

8.On AAA1 machine run command to check cluster is in AAA1 or AAA2.

# clustat

If cluster is in AA2 then run command on AAA1 machine

# clusvcadm –d Radius

#clusvcadm –e Radius –m `hostname`

Check by ping 192.168.10.104

9.Nitrobit 1 and Nitrobit 2 machine automatically started properly and not required to run any.

TROUBLESHOOTING WITH SERVERS

Note : If network fails then server restart automatically when machine is in cluster (HP-ILO restart server then they fence services to another server)
If network fails long time then it can’t able to fence services to another but then retry up to 10 minutes after still network is down then both the machines are shut-off automatically(HP-ILO has an capability to shutdown the machine).
After network is UP then machines are properly start according to above documentation.


DOMAIN SERVER :

1.If Domain server is shutoff then read the documentation above to start the domain servers.

2.If in case domain 1 machine is off and domain2 is in cluster and wanted to change the state of cluster

Run these command on Domain 2 firstly

# service rgmanager stop

# service cman stop

3.After on the Domain1 machine :

Run these command on Domain2

#ifup bridge0

4.After both the machine are in network and ON, then run command simultaneously on both the machine

#service cman start

#service rgmanager start

LDAP SERVER:

1.If LDAP server is not starting , when u restart the ldap service then error comes then run this command on both the server.

On LDAP1

# cp /etc/dirsrv/slapd-ldap1/dse.ldif.bak /etc/dirsrv/slapd-ldap1/dse.ldif

On LDAP2

#cp /etc/dirsrv/slapd-ldap2/dse.ldif.bak /etc/dirsrv/slapd-ldap2/dse.ldif

2.This command is run because dse.ldif file is corrupted when server unproperly shutoff/ fencing is not done.

3.If cluster is in Node2(LDAP2) and wanted to change the state then check Node1 server is ON or OFF.

4.If Node1 is OFF then ON the machine or ON then run these commands on Node1(LDAP1)

# clusvcadm –d dummy_httpd

# clusvcadm –e dummy_httpd –m `hostname`

Same process for Squid and AAA Server to change the state of cluster.