3PAR: configure AFC with SSDs only for AFC

When initialize a new 3PAR, depending on your protection level, at least the size of one disk per disk-type will be reserved as spare. Also when you add disks to a running system, spare-chunklets are created. This is done by admithw. For example when you initialize a system containing 2 SSDs for adaptive flash cache (AFC) the size of one SSD is reserved for spare. So without writing any byte to SSDs, 50% is already used! So how to be able to get the most space for AFC?

1. Check the capacity of you physical disks (PDs) by running showpd -c. You see free and used chunklet-counts. Also spare-chunklet are listed.
2. Remove spare-chunklets from disk by running: removespare n:a. n stands for PD-ID of your SSD.
3. Check again capacity by running showpd -c. You can see, free is the same as before, but uninit is the size of previous spares.
4. Wait a few minutes and you can see, uninit decreases and free increases. Chunklets get initialized automatically.
5. When all chunklets are initialized and therefore free, go ahead and create cache for AFC in GUI or CLI (createflashcache). SSMC needs a few minutes to see new free space on PDs.

Advertisements

“Download failed” when using proxy for VCSA update

There is still a bug in current VCSA 6.5 U1g. When you use a proxy server in management interface (Port 5480) to access VMware repository for updates, you get a “Download failed” error. The problem is that in configuration file https-proxy setting will not be set. To resolve the issue, do the following:

1. Log in to VCSA shell
2. Edit /etc/sysconfig/proxy
   • • set entry HTTPS_PROXY like HTTP_PROXY
     • example:
       • HTTP_PROXY="http://proxy:3128/"
       • HTTPS_PROXY="http://proxy:3128/"

    

No reboot or service restart is necessary. Just re-try to scan repository.

3PAR (iSCSI) – resolve high write latency on ESXi hosts

FC AFA (all flash array) 3PAR  systems show quite a good latency on reads and writes. When operating an iSCSI AFA 3PAR it could happen that the systems show a rather high write latency on ESXi hosts. In this post you can read how to fix this.

Update VCSA 6.5 with selected version

When you currently try to update your vCenter Service Appliance (VCSA) 6.5 in Virtual Appliance Management Interface (VAMI) (Port 5480) your only suggestion is Update 2. A new feature of VCSA 6.7 is the ability to select the update you want to install. In 6.5 the latest update will be installed when using default repository.

vSphere 6.5 U2 is a very uncommon update-version. First, there are not just fixes and updates in U2 but also new features (see here) that are also available in vSphere 6.7. Second, U2 gets available after 6.7, so it is not possible to upgrade 6.5 U2 to 6.7 GA. Therefore updating to U2 should be well considered.

What are your options to stay within a Update version respectively select your update version?

simple linux Script to shut down VMs and ESXi host

Here is a very simple linux bash script to shut down all VMs of a ESXi host and the host itself, for example when a power failure occurs, this script can be used in UPS software. Some time ago a posted how to use such script in an HPE UPS environment. You can find the post here.

Backup from snapshot on target 3PAR

In a 3PAR Peer Persistence configuration volumes are synchronously replicated between two 3PAR arrays. For each volume one array acts as primary or source array. This array exports the volume for read/write. The other array acts as secondary or target for replication. For performance reasons it could be an advantage when backup reads data from target array. This is easily possible with Veeam B&R 9.5.

How do IO Limits look like

In this blog post you can see how to configure IO limits and how ESXi hosts executes these limits. Furthermore you will see how IO size matters.