incus/dhcp: Explicitly requesting 51/58/59, derive sane T1, add jitter #2383
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
The built-in DHCPv4 client used for OCI containers could hammer the DHCP server with a renewal every 60 seconds. This happened when the initial DHCPOFFER lacked renewal/rebind timers (Options 58/59): the renewal loop slept using IPAddressRenewalTime(time.Minute), so the fallback value (1m) was used repeatedly. Additionally, our initial request implicitly asked for a fixed lease time (Option 51) and did not include a Parameter Request List for 51/58/59. Many servers respond by honoring the client’s requested lease (e.g. 3600s) and omitting 58/59 in the OFFER/ACK. Together these two behaviors produced one-minute renewals even when the server was configured for multi-hour leases. This change fixes the above by: 1. Not requesting a fixed lease in the client message (WithoutOption(OptionIPAddressLeaseTime)), allowing the server’s default-lease-time to apply. 2. Explicitly requesting critical options via PRL on both REQUEST and RENEW: 1,3,6,15,51,58,59 (and 121 if present). This makes servers include Lease-Time/T1/T2. 3. Deriving a sane T1 for the renewal sleep: prefer timers from the ACK, then OFFER; if still absent, use 50% of the offered lease-time (Option 51); finally fall back to 1m. Adding ±5% jitter to the renewal sleep to avoid herd effects when many containers renew at once. Signed-off-by: Andrey Mozharovsky <andrii.mozharovskyi@virtuozzo.com>
|
Thanks! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The built-in DHCPv4 client used for OCI containers can overwhelm the DHCP server with renewals every 60 seconds. This occurred when the initial DHCPOFFER lacked renewal/rebind timers (Options 58/59). The renewal loop slept using IPAddressRenewalTime(time.Minute), so the fallback value (1 minute) was used repeatedly.
Additionally, the initial request implicitly asked for a fixed lease time (Option 51) and did not include a Parameter Request List for 51/58/59. Many servers respond by honoring the client’s requested lease (e.g., 3600s) and omitting 58/59 in the OFFER/ACK. Together, these two behaviors produced one-minute renewals even when the server was configured for multi-hour leases.
This change fixes the above by:
Not requesting a fixed lease in the client message (WithoutOption(OptionIPAddressLeaseTime)), allowing the server’s default-lease-time to apply.
Explicitly requesting critical options via PRL on both REQUEST and RENEW: 1,3,6,15,51,58,59 (and 121 if present). This makes servers include Lease-Time/T1/T2.
Deriving a sane T1 for the renewal sleep: prefer timers from the ACK, then OFFER; if still absent, use 50% of the offered lease-time (Option 51); finally fall back to 1m.
Adding ±5% jitter to the renewal sleep to avoid herd effects when many containers renew at once.