Hello All,

Hope you are doing well!

1. What problem/issue/behavior are you having trouble with? What do you expect to see?

We have rhel 6.6 server and we we are seeing to much delay on server and we are unable to disable it on the server.

[root@Server~]# netstat -s|grep dela
249698 delayed acks sent
20 delayed acks further delayed because of locked socket
[root@Server~]#

Where are you experiencing the behavior? What environment?
We are experiencing while reading data, writing is fine but its too slow while reading the data on mounted EMC storage

Red Hat Enterprise Linux Server release 6.6 (Santiago)

We are testing reading and writing with dd command tool. Testing result as below

[root@Server ~]# dd of=/dev/null bs=1024k count=2000 if=/test1/testingfile_5g
2000+0 records in
2000+0 records out
2097152000 bytes (2.1 GB) copied, 212.389 s, 9.9 MB/s

[root@Server ~]# dd of=/dev/null bs=1024k count=2000 if=/test/testingfile_5g
2000+0 records in
2000+0 records out
2097152000 bytes (2.1 GB) copied, 61.1459 s, 34.3 MB/s
[root@Server ~]#

We have below EMC storage and their mount point where we performed the test.
/dev/mapper/mpathbp1 is /test
/dev/mapper/mpathcp1 is /test1

We have Contacted EMC support and they have asked us to disable TCP ack delay in rhel 6.6. We have found below details on rhel portal to disable tcp ack delay but it works only on rhel 7.2 or it comes with kernal-rt package which is for rhel5.
https://access.redhat.com/solutions/407743
https://access.redhat.com/solutions/1529183#

We don't have any application hosted on this server, we are going to use this server for Rack DB server. Currently we are testing storage performance before putting this in production.

Storage details : EMC VNX 5400 (Block only)
Block software Version: 05.33.006.5.102
----------
NIC details
07:00.0 Ethernet controller: Broadcom Corporation NetXtreme II BCM57810 10 Gigabit Ethernet (rev 10)

driver: bnx2x
version: 1.78.19-0
firmware-version: bc 7.12.83 phy 1.34
bus-info: 0000:07:00.0
supports-statistics: yes
supports-test: yes
supports-eeprom-access: yes
supports-register-dump: yes
supports-priv-flags: yes
-------------
04:00.0 Ethernet controller: Broadcom Corporation NetXtreme II BCM57810 10 Gigabit Ethernet (rev 10)
[root@Server~]# ethtool -i eth0
driver: bnx2x
version: 1.78.19-0
firmware-version: bc 7.12.83 phy 1.34
bus-info: 0000:04:00.0
supports-statistics: yes
supports-test: yes
supports-eeprom-access: yes
supports-register-dump: yes
supports-priv-flags: yes
[root@Server~]#


We don't have RHEL support with us hence looking for help.


Thanks
Sunil

For a production system running RedHat you should contact RedHat support.

1 members found this post helpful.

All you've really told us is you're using RHEL 6.6. You say NOTHING about what you're reading/writing, what application(s)/service(s) you're talking about, how the EMC is connected, through what kind of HBA, what drivers, or ANYTHING that would help us help you.
Since you need 'urgent' help, I suggest you contact both EMC and Red Hat, since you're in a supported environment, and you are PAYING FOR support from them both, RIGHT??? Asking for/expecting 'urgent' help from volunteers is fairly rude.

Thank you giving your valuable time for my issue.

I have edited my main post and given all required details.

Thank you very much

Editing your posts makes things harder for people, since it makes things less clear.
Then PAY FOR SUPPORT from RHEL, as you should. You'd then get the patch. And if this is a NEW server, stop where you are and THINK about what you're doing. You're putting a new system in, running an OLD version of the OS, and you're expecting good results....why?? Load the latest version of RHEL if you're PAYING for it, or load the free version of CentOS.

If you're using EMC storage, not paying for RHEL is just plain stupid. Your company OBVIOUSLY has the money to afford it. Pay for support, load the latest version, and use the support/patching you're entitled to. Otherwise, use the free CentOS software, and hope for the best.