How to connect to ESX Server using SSH (SFTP)
- System configuration of the test server
- Outdated ESX versions
- ESX server set up. Enabling SSH
- Establishing SSH connection and mounting VMFS disk in VMFS Recovery
- Recommendations, performance issues and hints
An ESX server v.6.5 is running as a guest inside a VM to ease testing, simulation and display various hardware configurations. There are two virtual hard drives provided to host ESX server: 20Gb and 10Gb. VMFS volume created by ESX resides on both disks combining set of two disks in a JBOD dynamic disk.
Please be advised that ESX server has got full support of SSH and SFTP since version ~5. Despite declared SSH and SFTP support in earlier ESX v.4.3 (SSH only) and ESX v.4.7 (SSH + SFTP), we recommend to update ESX 4.x version to ESX 5.0 before establishing SSH connection because of stability and performance issues.
By default SSH is disabled on ESX server because of security reasons. Therefore you need to enable it if you haven’t done it before.
To do it, open ESX server and press F2 to proceed to it’s options.
Enter user name and password to authenticate.
Select “Troubleshooting options” from the list and press Enter.
Check the status of SSH at the “Disable SSH” option and make necessary changes if required.
Now your ESX server is ready to accept SSH connections.
Return to the Windows PC where VMFS Recovery is installed, run it and go to "Drives->Connect to remote SSH" option. As an alternative, you can press CTRL+N on the keyboard.
Enter IP of ESX server to "Address [:port]" field, add port value after IP if you are using other port that is set by default. Default value for SSH port is 22.
Fill in "Username" and "Password" values and click "Connect" button to proceed.
If SSH connection to ESX server is successful you will see the following screen with mode selection:
Please note that you should select “Physical Drive” option for all data recovery purposes. Only a direct access to the physical drives allows unerasing VMDK files, reading corrupt disk sectors and reconstructing corrupt VMFS file system.
It is quite possible that your VMFS disk is a joined disk of several physical drives. In this case you need to run the scan on each physical drive that stores VMFS disk. On the screen above is exactly this case. VMFS volume is located on a JBOD disk that consists of 2 disks 20Gb and 10Gb.
If you are using a hardware RAID disk or iSCSI NAS storage, both physical drives will be displayed as a single large disk. If it’s RAID to be reconstructed, please refer to the corresponding section for more information.
Other two modes ("Datastores" and "Files") represent data on a same level or interaction with ESX server as Vsphere does. Therefore these options make sense to be used when Vsphere is not installed on the client PC or in similar cases.
Wait until the connection is established and you’ll be returned to the disk list and should see available VMFS partition(section "Hard drives"), physical drives from ESX server(section "Disk Images"), JBOD disk of two virtual drives that forms a disk for VMFS volume("RAID disk" section).
There are many ESX system volumes on the screen, that you can ignore.
All freshly added disks from RAID or SSH connection are marked red to be easily spotted. Red marks don’t mean any type of damage nor disk corruption.
To proceed to recovery of VMDK files, double click on the VMFS volume with lost data at the "Hard Drives" section. More detailed information about data recovery from VMFS disk can be found here.
We have explored all possible configurations and performed a lot of tests with SSH protocol in order to improve access speed. However, the best performance we’ve achieved is 30-35% available network bandwidth regardless of the Ethernet speed used. Professional SSH clients may be a couple of percent faster, but it’s still can not be considered as satisfying speed, we had to look for other options.iSCSI
Our tests have discovered that iSCSI protocol can achieve 50-60% of the available bandwidth, which makes it a much better choice for network recovery. Yet, we admit that iSCSI may be a more complicated in set up and have no out-of-box support in ESX server. Because of such a big difference compared to SSH, iSCSI is a recommended choice to recover large amounts of data. See a more detailed iSCSI setup guide here.Vsphere proprietary high speed protocols
There are several proprietary protocols built in Vsphere that allow significantly more speed with almost 100% available bandwidth covered. However these protocol access data on the disk on a file level and therefore it’s Vsphere’s driver to read disk. For any data recovery appliance, you need to provide direct disk access with RAW cluster data, and these protocols are not provide low level access and therefore can not be used for data recovery purposes.