etegro/opensan

opensan's Issues

Working per-package updates

Most of code's changes are related exclusively to interpreted Lua, dash,
Perl, HTML/CSS, LuCI-templates, JavaScript? code. Most of these changes
do not require system rebooting and overall system image rewriting.

OpenWRT provides out-of-box packages updating, using opkg subsystem. We
should check:

• If it is working for us
• Do we have corresponding packages versions incrementing
• Are package's dependencies fully correct and satisfied

Invalid LogicalVolume+Snapshots sort

Current LogicalVolume? with Snapshots sort implementation fully based upon special Snapshots naming scheme. If Snapshots are created manually, then they may appear everywhere in result matrix.

Incorrect hotspare availability calculations

We have issues where much smaller hard drive can be added (drop down menu is shown) to array consisting of bigger ones.

Also there are issues when hard drive can be successfuly added, but warning message about it's bigger size is shown (however hard drive's size is equal to all other array's members).

Snapshot's readonly flag

Determine is there any need in subject. LVM provides that possibility and it seems that read-only snapshots are very well suited for backups in most cases, without any need to think about permanent space incrementing.

Send bonding-related patches to upstream

Send everything related to network bonding to the upstream OpenWRT project, as it may seem useful for them.

Storing configuration directly on hard drives

Currently all configuration (except MD and LVM-related metainformation)
is stored via UCI subsystem on a bootable root flash device. It will be
much better to store it also on a hard drives directly. Other systems
called this feature "Configuration on Disks".

Need to understand how we can reach this target. Primarily the first aim
for checking is DDF (Disk Data Format) format that can include several
megabytes of user-defined data.

Failed physicals should not participate in comparision

When physical is going to be added to logical drive, JavaScript? scripts checks it's size and model. It checks if all drives inside logical drive are the same. If not, then it will warn user that identical models should be used.

If logical contains failed drive, then of course it's model will be empty and thus different from all other ones. And JS scripts will warn user. They must ignore failed drives without confusing user.

Extended logical drive add form

• Ability to choose stripe/RAID-level-block/chunk size
• Ability to choose RAID-level's algorithm to work with (such as left-assymetric, left-symmetric, la, ra, ls, rs, and so on for RAID5 level)

Also we have to check all those algorithms and option values and decide
what default values (for modern big and fast SAS/SATA) will we choose.
And do we need to provide users ability to change them at least.

Support for ETegro Hyperion RS230 G4 Backplane

Add ability for render front panel of Hyperion RS230 G4 in "SAN->Monitoring" tab.

AccessPattern's transport validation

LogicalVolume? can be "shared" amongst only exactly via single SCST's
transport (either FibreChannel? or iSCSI for example). However, nothing
denies binding LogicalVolume? to several AccessPatterns? (with common
transport).

DiskDataFormat support

Our Lua-written Einarc implementation must support DiskDataFormat? (DDF) by default. It is becoming the de-facto standard for most hardware and software RAID controllers.

please update root-zone trust-anchors in root.key file

Users of your software that enable DNSSEC will not be able to validate DNS after October the 11th 2018.

Your repository contains a root key file without the new DNSSEC trust-anchors:

. IN DS 19036 8 2 49AAC11D7B6F6446702E54A1607371607A1A41855200FD2CE1CDDE32F24E8FB5
It should also include:

. IN DS 20326 8 2 E06D44B80B8F1D39A95C0B0D7C65D08458E880409BBC683457104237C7F8EC8D

More information can be found at: https://www.icann.org/resources/pages/ksk-rollover

Please don’t hesitate to get in touch.

Warmly,

Roy Arends
ICANN

Separate MPT SAS kernel module package

Currently we have got dirty workaround to include MPT SAS kernel module
driver during the build process. It should be made as a separate
package.

HTTPS support

Think about how to enable HTTP support for administration through the
web-interface. How to deal with certificates, encryption options and the
small HTTP-server itself.

Real-time logical drive's initialization progress

Logical drive's initialization process is much better to be working in
real-time mode. For example, like it doned in LuCI's administration
web-interface via simple AJAX request.