Overview
| Comment: | Eliminated load delays of less than 1 second. Eliminated INFO when load is acceptable. |
|---|---|
| Downloads: | Tarball | ZIP archive | SQL archive |
| Timelines: | family | ancestors | descendants | both | v1.65-real |
| Files: | files | file ages | folders |
| SHA1: |
bd581f5fff38d1f72ac52eb21792e519 |
| User & Date: | mmgraham on 2021-02-12 16:44:51 |
| Other Links: | branch diff | manifest | tags |
Context
|
2021-02-14
| ||
| 19:54 | Bunch of minor fixes/cleanup check-in: 4aae235143 user: matt tags: v1.65-real | |
|
2021-02-12
| ||
| 16:44 | Eliminated load delays of less than 1 second. Eliminated INFO when load is acceptable. check-in: bd581f5fff user: mmgraham tags: v1.65-real | |
| 12:23 | Avoided server logs that have exited. Detected when a server is in dbprep, and wait 25 seconds for that to finish. check-in: bbf1632194 user: mmgraham tags: v1.65-real | |
Changes
Modified common.scm from [28bc361690] to [5dd52b20b3].
| ︙ | ︙ | |||
2185 2186 2187 2188 2189 2190 2191 | ;; effective load accounts for load jumps, this should elminate all the first-next-avg, adjwait, load-jump-limit ;; etc. (effective-load (common:get-intercept first next)) (recommended-delay (common:get-delay effective-load numcpus)) (effective-host (or remote-host "localhost")) (normalized-effective-load (/ effective-load numcpus)) (will-wait (> normalized-effective-load maxnormload))) | | | | > > | 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 |
;; effective load accounts for load jumps, this should elminate all the first-next-avg, adjwait, load-jump-limit
;; etc.
(effective-load (common:get-intercept first next))
(recommended-delay (common:get-delay effective-load numcpus))
(effective-host (or remote-host "localhost"))
(normalized-effective-load (/ effective-load numcpus))
(will-wait (> normalized-effective-load maxnormload)))
(if (> recommended-delay 1)
(let* ((actual-delay (min recommended-delay 30)))
(if (common:low-noise-print 30 (conc (round actual-delay) "-safe-load"))
(debug:print-info 0 *default-log-port* "Load control, delaying "
actual-delay " seconds to maintain safe load. current normalized effective load is "
normalized-effective-load". maxnormload = " maxnormload " numcpus = " numcpus " loadavg = " loadavg " effective-load = " effective-load))
(thread-sleep! actual-delay)))
(cond
;; bad data, try again to get the data
((not will-wait)
;; (if (common:low-noise-print 30 (conc (round normalized-effective-load) "-load-acceptable-" effective-host))
;; (debug:print 0 *default-log-port* "Effective load on " effective-host " is acceptable at " effective-load " continuing.")))
((and (< first 0) ;; this indicates the loadavg data is bad - machine may not be reachable
(> num-tries 0))
(debug:print 0 *default-log-port* "WARNING: received bad data from get-cpu-load "
first ", we'll sleep 10s and try " num-tries " more times.")
(thread-sleep! 10)
(common:wait-for-cpuload maxnormload numcpus-in
count: count remote-host: remote-host num-tries: (- num-tries 1)))
;; need to wait for load to drop
((and will-wait ;; (> first adjmaxload)
(> count 0))
(debug:print-info 0 *default-log-port*
"Delaying 15" ;; adjwait
" seconds due to normalized effective load " normalized-effective-load ;; first
" exceeding max of " adjmaxload
|
| ︙ | ︙ |