Professional Documents
Culture Documents
VPLEX Change Control Process For CCA - V1.2-May7 - 2015
VPLEX Change Control Process For CCA - V1.2-May7 - 2015
Version 1.2
May 7, 2015
Major Contributors: John Colburn, Martin Bowesman, David Bacon, Brian McGovern and Mike
Neiser
There are documents located on the VPLEX Procedure Generator and support.emc.com that the
submitter and the customer should be referring to when configuring a Host and Storage device
to be used with VPLEX to ensure they get the best performance from these devices and to avoid
issues that are pointed out in these documents.
When selecting any of the upgrade or convert options listed above, you will receive the
following message reminding you that you must submit a CCA for this activity.
You will not get this message when you click on “Install VPLEX”. While not mandatory, it is
recommended to submit a CCA for any new installation to be made aware of any known issues,
ETAs, and workarounds for the latest code level. No VPLEX logs are necessary for a VPLEX
installation CCA. See notes below for details on CCA file requirements.
The replacement of a component (FRU) does not require a CCA action as this is not an upgrade or
change to the VPLEX configuration.
Review the GeoSynchrony release notes with the customer to ensure all requirements have
been met before submitting the CCA to ensure a smooth process.
Do not enter tasks for management server upgrades on each cluster. CCA reviewers
only need to review the main cluster software upgrade tasks.
3. Multiple VPLEX Locals must be submitted as separate CCAs. Each CCA should have a
SINGLE task referencing the affected cluster. The SR referenced in that task must
reference an OPEN SR associated with the serial number of the affected cluster.
4. All SRs associated with the CCA MUST have an open TASK in the SR and this Task MUST
remain opened until the CCA, conditionally approved, task(s) have been completed. This
will ensure that the SR is not auto closed by the rule that checks SRs for Open Tasks and
if none are found the SR is auto closed.
7. Each CCA project must have a Switch_data.txt file attached which indicates the type of
Fibre Channel switches that each cluster is attached to and the version of code running
on that switch. See below for further explanation.
Your file attachments should look like this in CCA5 (with your appropriate C1_####.log
and C2_####.log files, along with the Switch_data.txt file):
8. The referenced upgrade SR(s) in the CCA must have the .tar file attached from running
“collect-diagnostics --noextended” on each cluster. This data is important to
have available in the event of problems during or after the NDU. The collect-diagnostic
logs MUST be recent logs, a day or two at most before the CCA is submitted. Please see
note below for more details about the file handling.
3. Once you log into the VPLEX CLI, run the command ‘ndu pre-check’
Note: If the setup is a Metro or Geo you need to run the ‘ndu pre-check’ on both
clusters.
o If there are any errors or warnings reported. Please discuss these with the
customer first and explain that all errors will need to be corrected before the
activity can proceed. These issues will most certainly be highlighted in the CCA
review.
o Common errors consist of:
1. Time drift between directors and Management servers. Follow
https://support.emc.com/kb/14180 to correct this.
2. Front-end and back-end connectivity issues:
a. Storage view HA issues (one port in each upgrader set, which is the
first upgraders, which are the ‘A’ directors and the second
upgraders which are the ‘B’ directors). Ndu pre-check will complain
but it’s normal on some configs to have storage views with only one
port on an ‘A’ and one on a ‘B’ director. Review the output to see if
the storage view names are repeated in the output. As long as each
view has a port on an ‘A’ and a ‘B’ director, it’s not an issue.
b. Over-pathing BE volumes. Inform the customer that they need to
reduce paths via the procedure found in SolVe.
3. Meta-volumes on single array or not backed up.
a. Single array errors are a common error when a single BE array is
available on a cluster. Meta-volume backup issues need to be
corrected via the troubleshooting info in the applicable upgrade
procedure.
4. The auto-resume attribute of the distributed device(s) is set to false. See
https://support.emc.com/kb/92664 for details.
4. After the initial errors are corrected (meta-volumes on the same BE array errors are
acceptable) and warnings are addressed, run the commands specified below. See
below for file handling instructions.
With a value of ‘0’ in the ‘exported volumes’ column, it means that this
storage view, ‘doc_example’ is empty. Therefore, in this example, no
action needs to be taken. However, if there are non-empty storage
views without a LUN 0, that issue must be corrected, by the customer,
prior to the upgrade.
Please consult the matrix in Table 1 to see the rules on how a CCA
Approver MUST proceed if there are VPLEX storage views missing a
LUN 0.
Note: For all above upgrade paths if NPIV is not in use and LUN 0 is present in all storage
views, no action is necessary. Note the LUN0 issue is fixed in upgrades from 5.3P1 to later
codes.
6. As noted above, if this is a VPLEX Metro or Geo, please ensure that you create a single
CCA with a task for each cluster. See below for file handling instructions for the CCA.
7. As noted above, the scheduled date for the upgrade should be no less than 10 days
after the date that you submit your project. We ask that the data collected for the CCA
be recent, as we want to ensure that there is minimal time between the review and the
upgrade. See Emergency CCA Notes.
8. Once you submit the CCA, please verify that the CCA has been properly submitted and
was not auto rejected for any reason.
9. When your CCA has been approved you will receive a notification from CCA5 indicating
the approval with conditions. You will also receive an email from the VPLEX CCA
10. As indicated in the upgrade procedures, please ensure that prior to the upgrade the
system status has not changed since the CCA was submitted. Always upgrade
according to the latest Procedure Generator output.
11. When closing the CCA, please ensure that call home is functioning appropriately and
that the IB is updated with the new configuration state.
CCA COMMANDS
Required Command outputs to capture from each VPLEX cluster are included in the embedded
“cca_commands.txt” file below.
cca_commands.txt
Use the embedded file to COPY and PASTE the commands into your putty sessions for each
cluster being upgraded. Please note that if the customer has change the cluster names from the
default of “cluster-1” and “cluster-2” you will need to modify those commands accordingly.
There are comments included in the embedded file that highlight the commands where this may
be an issue. Note “Appendix A” has the full listing of commands and comment also.
Once all commands have been run and ‘collect-diagnostics --noextended’ has been
run, please attach the captured output to CCA as per the naming conventions below. The .tar
file generated from running collect-diagnostics MUST be attached to the SR associated with the
cluster in the CCA. The CD collection is too large to upload to the CCA project/task.
If you are using a different SSH application to capture the output, save this output and attach it to the
CCA for review. Saving only “Printable output” creates a much cleaner log file for review.
What the base collect-diagnostic logs collected will look like in this folder:
The TLA = Top Level Assembly, is the serial number assigned to a complete VPLEX rack
setup.
It can be found on the sticker on the back of the rack or by typing one of the two
following commands,
VPlexcli:/> ls –t /engines/*::top-level-assembly
or
ll /engines/engine-<cluster-id>-<engine-id>
VPlexcli:/> ll /engines/engine-1-1
Once the ‘Start’ button is clicked, this would then split the 4.1Gb input file into 1.5Gb
segments that can be attached to the upgrade SR.
Along with the commands outputs collected from the VPLEX, you will also need to collect the
customer’s SAN switch type connected to the VPLEX and version of code running on those
switches . Please provide this detail in a text file named ‘Switch_data.txt’ attached to the
project within your CCA.
One file per project is enough. Not submitting a Switch_data.txt file will generate an
auto rejection of the CCA.
Note that the filename is CASE SENSITIVE: Switch_data.txt
Please do not attach the output of supportshow, supportsave or show tech-support.
We only require switch type and OS version:
Cisco/9513/NX-OS 5.2(8a)
Brocade/DCX/FOS 6.4.3b
All data captured logs, using PuTTY logging, are to use the following naming scheme as we are
working to automate the initial submission process via the CCA5 website.
NOTE: Remember collect only PRINTABLE output in putty, this makes the data easier to read
and therefore speed up the review of your CCA.
After naming the logging files appropriately, please attach them to their respective task in the
CCA. If the logs are not properly named or missing, the CCA will be auto rejected by the
automated CCA5 process.
When an entire VPLEX, or just one cluster, is to be relocated within the same data
center, using the same array(s) and host(s). You will still need to follow the same CCA
process as for an upgrade.
If the entire VPLEX or just one cluster is to be relocated to a different data center and a
new array(s) and host(s) will be used follow the same steps as the upgrade steps. There
will be a lot of data provided for this last action so you will need to allow a minimum of
three weeks for this particular CCA to ensure all the areas that will need to be addressed
in the planning are reviewed, discussed with the customer and fully understood. Any
missteps or missing info will be grounds for rejection of this CCA.
Installation of a VPLEX
When a new VPLEX is to be installed there will be no required logs for review. There will
be conditional text included in the approval email. Please be sure to review this data.
As detailed in the VPLEX Configuration Guide (on support.emc.com) it is recommended
that all new installs be upgraded to the latest GA version of the GeoSynchrony code,
prior to configuring the cluster(s). This is done by performing the steps found in the
‘Upgrading an Unconfigured System’ section in the VPLEX Configuration Guide. Also,
installs are not done by RCM but by Implementation personnel only. Important:
Because of auto reject rules in CCA you must create and upload dummy files to the
CCA project using the naming convention above.
If it is a totally dark site then a manual CCA can be done by a locally trained FSS via FSS
request. They must follow the latest Reviewer’s Guide and submit a complete CCA with
dummy files uploaded in order to maintain trackability of the event. The assigned FSS
can then approve with the appropriate notes.
It is recommended that if the customer or field would like to make use of the RCM team for an
upgrade that the RCM Team be contacted first, before submitting the CCA, to setup their
availability for doing the upgrade. Then once you have a date the RCM Team will be available
make this the scheduled date in the CCA then submit the CCA along with all the rest of the
requirements for submission.
RCM Group Landing Page: RCM; Email: RCM@emc.com; or Phone: +353 (0)21 494 5555 (ext.
25400 from Hopkinton).
RCM Notes:
When requesting the RCM team via their web app there is a section titled ‘Product specific
Information’ where the last field requests you to provide a ‘CCA Approved Documentation’. For
this field take a copy of the approval e-mail that is sent after a VPLEX CCA is approved, or save a
copy and attach this as the approval documentation. If you are using Microsoft’s IE8 or 9 you
will get an error message when trying to attach the file. Please use Chrome and/or Firefox with
the RCM portal so that you are able to attach the file and not get the error.
Engineering has been working on an escalation and has determined that the present
code level the customer is running on their VPLEX has a code bug and the fix is in the
level of code engineering states they should upgrade to in order to correct the issue on
the VPLEX. Most times engineering will recommend upgrading to the latest level of
code. If the latest level of code does not have the fix then engineering may have to write
a HOT FIX for this customer’s issue.
Note:
A code with a HOT FIX is specific to the customer it was written for and CANNOT be used on
any other customer’s VPLEX, not even the same customer if they have a second VPLEX setup
without express permission from engineering.
It has been determined from investigation by VPLEX Support that the customer has
encountered a known issue at the level of code they are running and the fix is in the
next or latest level of code.
When a customer informs you they want to upgrade their VPLEX code to the latest code level
and want to do it inside of the 10-day lead time.
You are to inform the customer this is not permitted due to the critical nature of VPLEX in their
virtualized storage environment. The CCA process has to be followed and there is a minimum
10-day lead time required. The 10-day requirement is to allow time for corrections should any
issues be discovered during the CCA review. See “Policy for CCA submission” back at the top of
the document.
In addition, please consult the Emergency CCA process document on the VPLEX Change Control
Wiki for more details surrounding the submission and handling of, Emergency VPLEX CCAs.
date -u
version -a
ls -t /management-server::cluster-ip-seed
configuration get-product-type
ls -t /engines/*::top-level-assembly
##check engine family VS1 is WildCat, VS2 is Argonaut. Be aware of upgrade paths
ls -t /engines/**::engine-family
ll /engines/*/directors
director uptime
director appstatus
cluster summary
cluster status
ll /clusters/**/storage-arrays
virtual-volume summary
storage-volume summary
extent summary
local-device summary
connectivity show
connectivity validate-be
ll /engines/**/ports
ll /engines/**/sfps
ll /clusters/*/exports/initiator-ports
ll /engines/**/stand-by-power-supplies/**/
##METRO or GEO
ds summary
consistency-group summary
ll /clusters/**/consistency-groups/
vpn status
connectivity validate-wan-com
ll /
ll /cluster-witness
ll /cluster-witness/components
##Recoverpoint##
ll /recoverpoint
ll /recoverpoint/rpa-clusters
health-check
##LUN0 check
## If clusters have non-default names, substitute them in place of
## cluster-1 and cluster-2 in the following commands
## There must NOT be any non-empty storage views where the NEXT free lun is 0, this
must be fixed prior to NDU
ll /clusters
export storage-view find --free-lun -c cluster-1
export storage-view find --free-lun -c cluster-2
version -a --verbose
ll /monitoring/directors/*/monitors
snmp-agent status
ndu pre-check --verbose
collect-diagnostics --noextended
##Attach the resultant .tar.gz file to the SR for this Cluster.