Releases: maxiepax/go-via
v0.4.5
-
Added ability to override the ks.cfg file on both Group and Host level.
The order of precedence is Host -> Group -> Default.
Any options found at https://docs.vmware.com/en/VMware-vSphere/7.0/com.vmware.esxi.upgrade.doc/GUID-61A14EBB-5CF3-43EE-87EF-DB8EC6D83698.html can be used in the ks.cfg file. -
Upgraded a number of angular packages that had security exploits.
-
Hosts can now be edited.
v0.4.3
BREAKING:
Due to terraform i had to move "bootdisk" out from the Options structure, it's now sideby side with vlan, dns, ntp etc. Sorry about this :/
Bugs:
fixed issue where GUI wouldn't render properly when a host was created by API.
New:
added support for Terraform Provider https://github.com/maxiepax/terraform-provider-govia
v0.4.2
- moved a lot of the post-config code into individual functions and sorted proper error handling
- first version of the callback feature implemented. add a callback url to Group settings to have go-via callback to a url after host has completed postconfig. URL must accept HTTP-POST, and payload is json string. The string is a combination of "address","group","pool" with sensitive information redacted.
v0.4.1
v0.4.0
- Added support for arm_64 architecture. This is completely transparent to the user, simply upload an arm_64 build of ESXi and it will install it.
- Project Monterey (SmartNIC) support added. Validated card is NVIDIA BlueField-2. A requirement is to specify mpx device option ( mpx.vmhba64:C0:T0:L0 for BF2 eMMC ).
- Added safeguards in both UI and API to not be able to remove Groups, Pools or Images that are currently in use.
- Added syslog firewall configuration. Before it would just configure syslog but since firewall is default closed for outbound syslog traffic, nothing would reach the syslog server.
- Fixed so that servers using nic other than vmnic0 will now work. both boot.cfg and ks.cfg will now specify to use vnmic of the mac-address defined to boot from as nic to add to vSwitch0.
- Added 'Create VMFS' option that will consume the remaining space of the boot drive and create 'datastore1'.
v0.3.22
v0.3.21
Improvements:
- Added option to configure vmk0 and port-group 'VM Network' with a VLAN. Please ensure that after imaging the host will be reachable over that vlan since post-config is done via API.
- Re-modeld the Add/Update Groups modal so that it makes more sense and looks cleaner. There's no checkbox for things like "NTP", simply inputting NTP config in the input will now also automatically configure it.
Bugs:
- When starting from fresh a 'secrets.key' file was generated. This however was incorrectly loaded and caused an error when adding a Group, since it couldn't encrypt the Password being entered. Simply re-starting the application/container solved this since during startup it would detect an already .key and load it correctly. This is now fixed so that .key is loaded properly even during first-run.
v0.3.20
Bug fixes:
- when adding a mac address with formats AA:BB:CC:11:22:33 or AA-BB-CC-11-22-33 host would not get dhcp offer. all mac-address formats are now unified in backend to lowercase aa:bb:cc:11:22:33.
- when changing settings in Edit mode of Groups the settings would take effect but the GUI would not update. GUI now updates this live.
- Editing a group and not changing the password would re-encrypt the encrypted password causing an error when ks.cfg was read by host, saying the password is longer than 41 characters. Password field is now left empty and if you change anything in the group and dont supply a new password, the password is kept the same.
Enhancements:
- Version and commit is now displayed at the top left corner to make it easier to know if you're running the latest version.
- A description field is now available when uploading new images. This allows you to enter information about drivers that may have been injected or other special things with the image.
- default landing page is now the dhcp pools instead of a blank page.
v0.3.10
- Option to generate certificate that will replace the VMware ESXi installer certificate with a cert that has valid CN and SAN fields.
This is a requirement to be able to provision hosts with Cloud Foundation. Please note that due to the fact that there is no way to via API restart the hostd service, an additional reboot is performed.
v0.3.00
- secrets added, all sensitive information is now encrypted and stored in database using AES-256. If you don't have AES-256 key one will be created in /secret/secret.key. Its important this key is kept from users who have shell access since they will be able to decrypt passwords etc with it.
IMPORTANT: After you have upgraded, you need to go into each Group, and re-type the password you want to encrypt the password using the key!
If you are using docker, please use the new docker-compose file.
- Passwords for groups/hosts are now stored encrypted in the database using AES-256.
- Added syslog as option for groups. syslog is added as a comma separated string, checking checkbox will configure syslog, enable service and restart it.
- Added option for "allowLegacyCPU" for those with older baremetal hosts that support UEFI but have CPUs that are unsupported with vSphere (for lab usage only).
- timer for postConfig has been reworked and is now stable and predictive. Default wait time between attempts to reach SOAP API is 10 seconds, and 360 retries are made. This allows bare metal hosts with huge amounts of memory sufficient amount of time to boot.
- DNS configuration actually only configured the search domain, now the domain portion of the FQDN. This has now been altered so that both domain and search domain are populated during postconfig.
- It's now possible to stop a re-image request via HTTP-REST API, and by pressing Cancel in the UI. This does not halt a ongoing re-image, but ensures that if you reboot the host and have it configured to attempt PXE, the DHCPd wont repsond to the request.
- ks.cfg --clearpart worked fine on VMs, but didn't work well on baremetal servers, have replaced with --clearpart --alldrives which seems to work a lot better.
- ks.cfg --firstdisk worked fine on VMs, but didn't work well on baremetal servers, replaced with --firstdisk="localesx,usb,ahci,vmw_ahci,VMware"
- Added HTTP-REST endpoint to manually trigger postConfig for occasions where postConfig timer timed out and you dont want to retry entire re-image process (and for more rapid development). use /v1/postconfig/$hostid to trigger.
- Removed lots of spew.Dump() and console.log() that is no longer required since process is now very stable.
- You can no longer add an IP to a subnet that is outside the scope of the subnet.