## root / Ganeti / HTools / Cluster.hs @ 234d8af0

History | View | Annotate | Download (30.5 kB)

1 |
{-| Implementation of cluster-wide logic. |
---|---|

2 | |

3 |
This module holds all pure cluster-logic; I\/O related functionality |

4 |
goes into the "Main" module for the individual binaries. |

5 | |

6 |
-} |

7 | |

8 |
module Ganeti.HTools.Cluster |

9 |
( |

10 |
-- * Types |

11 |
NodeList |

12 |
, InstanceList |

13 |
, NameList |

14 |
, Placement |

15 |
, Solution(..) |

16 |
, Table(..) |

17 |
, Removal |

18 |
, Score |

19 |
-- * Generic functions |

20 |
, totalResources |

21 |
-- * First phase functions |

22 |
, computeBadItems |

23 |
-- * Second phase functions |

24 |
, computeSolution |

25 |
, applySolution |

26 |
, printSolution |

27 |
, printSolutionLine |

28 |
, formatCmds |

29 |
, printNodes |

30 |
-- * Balacing functions |

31 |
, checkMove |

32 |
, compCV |

33 |
, printStats |

34 |
-- * Loading functions |

35 |
, loadData |

36 |
, checkData |

37 |
) where |

38 | |

39 |
import Data.List |

40 |
import Data.Maybe (isNothing, fromJust) |

41 |
import Text.Printf (printf) |

42 |
import Data.Function |

43 | |

44 |
import qualified Ganeti.HTools.Container as Container |

45 |
import qualified Ganeti.HTools.Instance as Instance |

46 |
import qualified Ganeti.HTools.Node as Node |

47 |
import Ganeti.HTools.Utils |

48 | |

49 |
type NodeList = Container.Container Node.Node |

50 |
type InstanceList = Container.Container Instance.Instance |

51 |
-- | The type used to hold idx-to-name mappings |

52 |
type NameList = [(Int, String)] |

53 |
-- | A separate name for the cluster score type |

54 |
type Score = Double |

55 | |

56 |
-- | The description of an instance placement. |

57 |
type Placement = (Int, Int, Int, Score) |

58 | |

59 |
{- | A cluster solution described as the solution delta and the list |

60 |
of placements. |

61 | |

62 |
-} |

63 |
data Solution = Solution Int [Placement] |

64 |
deriving (Eq, Ord, Show) |

65 | |

66 |
-- | Returns the delta of a solution or -1 for Nothing |

67 |
solutionDelta :: Maybe Solution -> Int |

68 |
solutionDelta sol = case sol of |

69 |
Just (Solution d _) -> d |

70 |
_ -> -1 |

71 | |

72 |
-- | A removal set. |

73 |
data Removal = Removal NodeList [Instance.Instance] |

74 | |

75 |
-- | An instance move definition |

76 |
data IMove = Failover -- ^ Failover the instance (f) |

77 |
| ReplacePrimary Int -- ^ Replace primary (f, r:np, f) |

78 |
| ReplaceSecondary Int -- ^ Replace secondary (r:ns) |

79 |
| ReplaceAndFailover Int -- ^ Replace secondary, failover (r:np, f) |

80 |
| FailoverAndReplace Int -- ^ Failover, replace secondary (f, r:ns) |

81 |
deriving (Show) |

82 | |

83 |
-- | The complete state for the balancing solution |

84 |
data Table = Table NodeList InstanceList Score [Placement] |

85 |
deriving (Show) |

86 | |

87 |
-- General functions |

88 | |

89 |
-- | Cap the removal list if needed. |

90 |
capRemovals :: [a] -> Int -> [a] |

91 |
capRemovals removals max_removals = |

92 |
if max_removals > 0 then |

93 |
take max_removals removals |

94 |
else |

95 |
removals |

96 | |

97 |
-- | Check if the given node list fails the N+1 check. |

98 |
verifyN1Check :: [Node.Node] -> Bool |

99 |
verifyN1Check nl = any Node.failN1 nl |

100 | |

101 |
-- | Verifies the N+1 status and return the affected nodes. |

102 |
verifyN1 :: [Node.Node] -> [Node.Node] |

103 |
verifyN1 nl = filter Node.failN1 nl |

104 | |

105 |
{-| Add an instance and return the new node and instance maps. -} |

106 |
addInstance :: NodeList -> Instance.Instance -> |

107 |
Node.Node -> Node.Node -> Maybe NodeList |

108 |
addInstance nl idata pri sec = |

109 |
let pdx = Node.idx pri |

110 |
sdx = Node.idx sec |

111 |
in do |

112 |
pnode <- Node.addPri pri idata |

113 |
snode <- Node.addSec sec idata pdx |

114 |
new_nl <- return $ Container.addTwo sdx snode |

115 |
pdx pnode nl |

116 |
return new_nl |

117 | |

118 |
-- | Remove an instance and return the new node and instance maps. |

119 |
removeInstance :: NodeList -> Instance.Instance -> NodeList |

120 |
removeInstance nl idata = |

121 |
let pnode = Instance.pnode idata |

122 |
snode = Instance.snode idata |

123 |
pn = Container.find pnode nl |

124 |
sn = Container.find snode nl |

125 |
new_nl = Container.addTwo |

126 |
pnode (Node.removePri pn idata) |

127 |
snode (Node.removeSec sn idata) nl in |

128 |
new_nl |

129 | |

130 |
-- | Remove an instance and return the new node map. |

131 |
removeInstances :: NodeList -> [Instance.Instance] -> NodeList |

132 |
removeInstances = foldl' removeInstance |

133 | |

134 |
-- | Compute the total free disk and memory in the cluster. |

135 |
totalResources :: Container.Container Node.Node -> (Int, Int) |

136 |
totalResources nl = |

137 |
foldl' |

138 |
(\ (mem, dsk) node -> (mem + (Node.f_mem node), |

139 |
dsk + (Node.f_dsk node))) |

140 |
(0, 0) (Container.elems nl) |

141 | |

142 |
{- | Compute a new version of a cluster given a solution. |

143 | |

144 |
This is not used for computing the solutions, but for applying a |

145 |
(known-good) solution to the original cluster for final display. |

146 | |

147 |
It first removes the relocated instances after which it places them on |

148 |
their new nodes. |

149 | |

150 |
-} |

151 |
applySolution :: NodeList -> InstanceList -> [Placement] -> NodeList |

152 |
applySolution nl il sol = |

153 |
let odxes = map (\ (a, b, c, _) -> (Container.find a il, |

154 |
Node.idx (Container.find b nl), |

155 |
Node.idx (Container.find c nl)) |

156 |
) sol |

157 |
idxes = (\ (x, _, _) -> x) (unzip3 odxes) |

158 |
nc = removeInstances nl idxes |

159 |
in |

160 |
foldl' (\ nz (a, b, c) -> |

161 |
let new_p = Container.find b nz |

162 |
new_s = Container.find c nz in |

163 |
fromJust (addInstance nz a new_p new_s) |

164 |
) nc odxes |

165 | |

166 | |

167 |
-- First phase functions |

168 | |

169 |
{- | Given a list 1,2,3..n build a list of pairs [(1, [2..n]), (2, |

170 |
[3..n]), ...] |

171 | |

172 |
-} |

173 |
genParts :: [a] -> Int -> [(a, [a])] |

174 |
genParts l count = |

175 |
case l of |

176 |
[] -> [] |

177 |
x:xs -> |

178 |
if length l < count then |

179 |
[] |

180 |
else |

181 |
(x, xs) : (genParts xs count) |

182 | |

183 |
-- | Generates combinations of count items from the names list. |

184 |
genNames :: Int -> [b] -> [[b]] |

185 |
genNames count1 names1 = |

186 |
let aux_fn count names current = |

187 |
case count of |

188 |
0 -> [current] |

189 |
_ -> |

190 |
concatMap |

191 |
(\ (x, xs) -> aux_fn (count - 1) xs (x:current)) |

192 |
(genParts names count) |

193 |
in |

194 |
aux_fn count1 names1 [] |

195 | |

196 |
{- | Computes the pair of bad nodes and instances. |

197 | |

198 |
The bad node list is computed via a simple 'verifyN1' check, and the |

199 |
bad instance list is the list of primary and secondary instances of |

200 |
those nodes. |

201 | |

202 |
-} |

203 |
computeBadItems :: NodeList -> InstanceList -> |

204 |
([Node.Node], [Instance.Instance]) |

205 |
computeBadItems nl il = |

206 |
let bad_nodes = verifyN1 $ filter (not . Node.offline) $ Container.elems nl |

207 |
bad_instances = map (\idx -> Container.find idx il) $ |

208 |
sort $ nub $ concat $ |

209 |
map (\ n -> (Node.slist n) ++ (Node.plist n)) bad_nodes |

210 |
in |

211 |
(bad_nodes, bad_instances) |

212 | |

213 | |

214 |
{- | Checks if removal of instances results in N+1 pass. |

215 | |

216 |
Note: the check removal cannot optimize by scanning only the affected |

217 |
nodes, since the cluster is known to be not healthy; only the check |

218 |
placement can make this shortcut. |

219 | |

220 |
-} |

221 |
checkRemoval :: NodeList -> [Instance.Instance] -> Maybe Removal |

222 |
checkRemoval nl victims = |

223 |
let nx = removeInstances nl victims |

224 |
failN1 = verifyN1Check (Container.elems nx) |

225 |
in |

226 |
if failN1 then |

227 |
Nothing |

228 |
else |

229 |
Just $ Removal nx victims |

230 | |

231 | |

232 |
-- | Computes the removals list for a given depth |

233 |
computeRemovals :: NodeList |

234 |
-> [Instance.Instance] |

235 |
-> Int |

236 |
-> [Maybe Removal] |

237 |
computeRemovals nl bad_instances depth = |

238 |
map (checkRemoval nl) $ genNames depth bad_instances |

239 | |

240 |
-- Second phase functions |

241 | |

242 |
-- | Single-node relocation cost |

243 |
nodeDelta :: Int -> Int -> Int -> Int |

244 |
nodeDelta i p s = |

245 |
if i == p || i == s then |

246 |
0 |

247 |
else |

248 |
1 |

249 | |

250 |
{-| Compute best solution. |

251 | |

252 |
This function compares two solutions, choosing the minimum valid |

253 |
solution. |

254 |
-} |

255 |
compareSolutions :: Maybe Solution -> Maybe Solution -> Maybe Solution |

256 |
compareSolutions a b = case (a, b) of |

257 |
(Nothing, x) -> x |

258 |
(x, Nothing) -> x |

259 |
(x, y) -> min x y |

260 | |

261 |
-- | Compute best table. Note that the ordering of the arguments is important. |

262 |
compareTables :: Table -> Table -> Table |

263 |
compareTables a@(Table _ _ a_cv _) b@(Table _ _ b_cv _ ) = |

264 |
if a_cv > b_cv then b else a |

265 | |

266 |
-- | Check if a given delta is worse then an existing solution. |

267 |
tooHighDelta :: Maybe Solution -> Int -> Int -> Bool |

268 |
tooHighDelta sol new_delta max_delta = |

269 |
if new_delta > max_delta && max_delta >=0 then |

270 |
True |

271 |
else |

272 |
case sol of |

273 |
Nothing -> False |

274 |
Just (Solution old_delta _) -> old_delta <= new_delta |

275 | |

276 |
{-| Check if placement of instances still keeps the cluster N+1 compliant. |

277 | |

278 |
This is the workhorse of the allocation algorithm: given the |

279 |
current node and instance maps, the list of instances to be |

280 |
placed, and the current solution, this will return all possible |

281 |
solution by recursing until all target instances are placed. |

282 | |

283 |
-} |

284 |
checkPlacement :: NodeList -- ^ The current node list |

285 |
-> [Instance.Instance] -- ^ List of instances still to place |

286 |
-> [Placement] -- ^ Partial solution until now |

287 |
-> Int -- ^ The delta of the partial solution |

288 |
-> Maybe Solution -- ^ The previous solution |

289 |
-> Int -- ^ Abort if the we go above this delta |

290 |
-> Maybe Solution -- ^ The new solution |

291 |
checkPlacement nl victims current current_delta prev_sol max_delta = |

292 |
let target = head victims |

293 |
opdx = Instance.pnode target |

294 |
osdx = Instance.snode target |

295 |
vtail = tail victims |

296 |
have_tail = (length vtail) > 0 |

297 |
nodes = Container.elems nl |

298 |
iidx = Instance.idx target |

299 |
in |

300 |
foldl' |

301 |
(\ accu_p pri -> |

302 |
let |

303 |
pri_idx = Node.idx pri |

304 |
upri_delta = current_delta + nodeDelta pri_idx opdx osdx |

305 |
new_pri = Node.addPri pri target |

306 |
fail_delta1 = tooHighDelta accu_p upri_delta max_delta |

307 |
in |

308 |
if fail_delta1 || isNothing(new_pri) then accu_p |

309 |
else let pri_nl = Container.add pri_idx (fromJust new_pri) nl in |

310 |
foldl' |

311 |
(\ accu sec -> |

312 |
let |

313 |
sec_idx = Node.idx sec |

314 |
upd_delta = upri_delta + |

315 |
nodeDelta sec_idx opdx osdx |

316 |
fail_delta2 = tooHighDelta accu upd_delta max_delta |

317 |
new_sec = Node.addSec sec target pri_idx |

318 |
in |

319 |
if sec_idx == pri_idx || fail_delta2 || |

320 |
isNothing new_sec then accu |

321 |
else let |

322 |
nx = Container.add sec_idx (fromJust new_sec) pri_nl |

323 |
upd_cv = compCV nx |

324 |
plc = (iidx, pri_idx, sec_idx, upd_cv) |

325 |
c2 = plc:current |

326 |
result = |

327 |
if have_tail then |

328 |
checkPlacement nx vtail c2 upd_delta |

329 |
accu max_delta |

330 |
else |

331 |
Just (Solution upd_delta c2) |

332 |
in compareSolutions accu result |

333 |
) accu_p nodes |

334 |
) prev_sol nodes |

335 | |

336 |
-- | Apply a move |

337 |
applyMove :: NodeList -> Instance.Instance |

338 |
-> IMove -> (Maybe NodeList, Instance.Instance, Int, Int) |

339 |
-- Failover (f) |

340 |
applyMove nl inst Failover = |

341 |
let old_pdx = Instance.pnode inst |

342 |
old_sdx = Instance.snode inst |

343 |
old_p = Container.find old_pdx nl |

344 |
old_s = Container.find old_sdx nl |

345 |
int_p = Node.removePri old_p inst |

346 |
int_s = Node.removeSec old_s inst |

347 |
new_nl = do -- Maybe monad |

348 |
new_p <- Node.addPri int_s inst |

349 |
new_s <- Node.addSec int_p inst old_sdx |

350 |
return $ Container.addTwo old_pdx new_s old_sdx new_p nl |

351 |
in (new_nl, Instance.setBoth inst old_sdx old_pdx, old_sdx, old_pdx) |

352 | |

353 |
-- Replace the primary (f:, r:np, f) |

354 |
applyMove nl inst (ReplacePrimary new_pdx) = |

355 |
let old_pdx = Instance.pnode inst |

356 |
old_sdx = Instance.snode inst |

357 |
old_p = Container.find old_pdx nl |

358 |
old_s = Container.find old_sdx nl |

359 |
tgt_n = Container.find new_pdx nl |

360 |
int_p = Node.removePri old_p inst |

361 |
int_s = Node.removeSec old_s inst |

362 |
new_nl = do -- Maybe monad |

363 |
new_p <- Node.addPri tgt_n inst |

364 |
new_s <- Node.addSec int_s inst new_pdx |

365 |
return $ Container.add new_pdx new_p $ |

366 |
Container.addTwo old_pdx int_p old_sdx new_s nl |

367 |
in (new_nl, Instance.setPri inst new_pdx, new_pdx, old_sdx) |

368 | |

369 |
-- Replace the secondary (r:ns) |

370 |
applyMove nl inst (ReplaceSecondary new_sdx) = |

371 |
let old_pdx = Instance.pnode inst |

372 |
old_sdx = Instance.snode inst |

373 |
old_s = Container.find old_sdx nl |

374 |
tgt_n = Container.find new_sdx nl |

375 |
int_s = Node.removeSec old_s inst |

376 |
new_nl = Node.addSec tgt_n inst old_pdx >>= |

377 |
\new_s -> return $ Container.addTwo new_sdx |

378 |
new_s old_sdx int_s nl |

379 |
in (new_nl, Instance.setSec inst new_sdx, old_pdx, new_sdx) |

380 | |

381 |
-- Replace the secondary and failover (r:np, f) |

382 |
applyMove nl inst (ReplaceAndFailover new_pdx) = |

383 |
let old_pdx = Instance.pnode inst |

384 |
old_sdx = Instance.snode inst |

385 |
old_p = Container.find old_pdx nl |

386 |
old_s = Container.find old_sdx nl |

387 |
tgt_n = Container.find new_pdx nl |

388 |
int_p = Node.removePri old_p inst |

389 |
int_s = Node.removeSec old_s inst |

390 |
new_nl = do -- Maybe monad |

391 |
new_p <- Node.addPri tgt_n inst |

392 |
new_s <- Node.addSec int_p inst new_pdx |

393 |
return $ Container.add new_pdx new_p $ |

394 |
Container.addTwo old_pdx new_s old_sdx int_s nl |

395 |
in (new_nl, Instance.setBoth inst new_pdx old_pdx, new_pdx, old_pdx) |

396 | |

397 |
-- Failver and replace the secondary (f, r:ns) |

398 |
applyMove nl inst (FailoverAndReplace new_sdx) = |

399 |
let old_pdx = Instance.pnode inst |

400 |
old_sdx = Instance.snode inst |

401 |
old_p = Container.find old_pdx nl |

402 |
old_s = Container.find old_sdx nl |

403 |
tgt_n = Container.find new_sdx nl |

404 |
int_p = Node.removePri old_p inst |

405 |
int_s = Node.removeSec old_s inst |

406 |
new_nl = do -- Maybe monad |

407 |
new_p <- Node.addPri int_s inst |

408 |
new_s <- Node.addSec tgt_n inst old_sdx |

409 |
return $ Container.add new_sdx new_s $ |

410 |
Container.addTwo old_sdx new_p old_pdx int_p nl |

411 |
in (new_nl, Instance.setBoth inst old_sdx new_sdx, old_sdx, new_sdx) |

412 | |

413 |
checkSingleStep :: Table -- ^ The original table |

414 |
-> Instance.Instance -- ^ The instance to move |

415 |
-> Table -- ^ The current best table |

416 |
-> IMove -- ^ The move to apply |

417 |
-> Table -- ^ The final best table |

418 |
checkSingleStep ini_tbl target cur_tbl move = |

419 |
let |

420 |
Table ini_nl ini_il _ ini_plc = ini_tbl |

421 |
(tmp_nl, new_inst, pri_idx, sec_idx) = applyMove ini_nl target move |

422 |
in |

423 |
if isNothing tmp_nl then cur_tbl |

424 |
else |

425 |
let tgt_idx = Instance.idx target |

426 |
upd_nl = fromJust tmp_nl |

427 |
upd_cvar = compCV upd_nl |

428 |
upd_il = Container.add tgt_idx new_inst ini_il |

429 |
upd_plc = (tgt_idx, pri_idx, sec_idx, upd_cvar):ini_plc |

430 |
upd_tbl = Table upd_nl upd_il upd_cvar upd_plc |

431 |
in |

432 |
compareTables cur_tbl upd_tbl |

433 | |

434 |
-- | Given the status of the current secondary as a valid new node |

435 |
-- and the current candidate target node, |

436 |
-- generate the possible moves for a instance. |

437 |
possibleMoves :: Bool -> Int -> [IMove] |

438 |
possibleMoves True tdx = |

439 |
[ReplaceSecondary tdx, |

440 |
ReplaceAndFailover tdx, |

441 |
ReplacePrimary tdx, |

442 |
FailoverAndReplace tdx] |

443 | |

444 |
possibleMoves False tdx = |

445 |
[ReplaceSecondary tdx, |

446 |
ReplaceAndFailover tdx] |

447 | |

448 |
-- | Compute the best move for a given instance. |

449 |
checkInstanceMove :: [Int] -- Allowed target node indices |

450 |
-> Table -- Original table |

451 |
-> Instance.Instance -- Instance to move |

452 |
-> Table -- Best new table for this instance |

453 |
checkInstanceMove nodes_idx ini_tbl target = |

454 |
let |

455 |
opdx = Instance.pnode target |

456 |
osdx = Instance.snode target |

457 |
nodes = filter (\idx -> idx /= opdx && idx /= osdx) nodes_idx |

458 |
use_secondary = elem osdx nodes_idx |

459 |
aft_failover = if use_secondary -- if allowed to failover |

460 |
then checkSingleStep ini_tbl target ini_tbl Failover |

461 |
else ini_tbl |

462 |
all_moves = concatMap (possibleMoves use_secondary) nodes |

463 |
in |

464 |
-- iterate over the possible nodes for this instance |

465 |
foldl' (checkSingleStep ini_tbl target) aft_failover all_moves |

466 | |

467 |
-- | Compute the best next move. |

468 |
checkMove :: [Int] -- ^ Allowed target node indices |

469 |
-> Table -- ^ The current solution |

470 |
-> [Instance.Instance] -- ^ List of instances still to move |

471 |
-> Table -- ^ The new solution |

472 |
checkMove nodes_idx ini_tbl victims = |

473 |
let Table _ _ _ ini_plc = ini_tbl |

474 |
-- iterate over all instances, computing the best move |

475 |
best_tbl = |

476 |
foldl' |

477 |
(\ step_tbl elem -> compareTables step_tbl $ |

478 |
checkInstanceMove nodes_idx ini_tbl elem) |

479 |
ini_tbl victims |

480 |
Table _ _ _ best_plc = best_tbl |

481 |
in |

482 |
if length best_plc == length ini_plc then -- no advancement |

483 |
ini_tbl |

484 |
else |

485 |
best_tbl |

486 | |

487 |
{- | Auxiliary function for solution computation. |

488 | |

489 |
We write this in an explicit recursive fashion in order to control |

490 |
early-abort in case we have met the min delta. We can't use foldr |

491 |
instead of explicit recursion since we need the accumulator for the |

492 |
abort decision. |

493 | |

494 |
-} |

495 |
advanceSolution :: [Maybe Removal] -- ^ The removal to process |

496 |
-> Int -- ^ Minimum delta parameter |

497 |
-> Int -- ^ Maximum delta parameter |

498 |
-> Maybe Solution -- ^ Current best solution |

499 |
-> Maybe Solution -- ^ New best solution |

500 |
advanceSolution [] _ _ sol = sol |

501 |
advanceSolution (Nothing:xs) m n sol = advanceSolution xs m n sol |

502 |
advanceSolution ((Just (Removal nx removed)):xs) min_d max_d prev_sol = |

503 |
let new_sol = checkPlacement nx removed [] 0 prev_sol max_d |

504 |
new_delta = solutionDelta $! new_sol |

505 |
in |

506 |
if new_delta >= 0 && new_delta <= min_d then |

507 |
new_sol |

508 |
else |

509 |
advanceSolution xs min_d max_d new_sol |

510 | |

511 |
-- | Computes the placement solution. |

512 |
solutionFromRemovals :: [Maybe Removal] -- ^ The list of (possible) removals |

513 |
-> Int -- ^ Minimum delta parameter |

514 |
-> Int -- ^ Maximum delta parameter |

515 |
-> Maybe Solution -- ^ The best solution found |

516 |
solutionFromRemovals removals min_delta max_delta = |

517 |
advanceSolution removals min_delta max_delta Nothing |

518 | |

519 |
{- | Computes the solution at the given depth. |

520 | |

521 |
This is a wrapper over both computeRemovals and |

522 |
solutionFromRemovals. In case we have no solution, we return Nothing. |

523 | |

524 |
-} |

525 |
computeSolution :: NodeList -- ^ The original node data |

526 |
-> [Instance.Instance] -- ^ The list of /bad/ instances |

527 |
-> Int -- ^ The /depth/ of removals |

528 |
-> Int -- ^ Maximum number of removals to process |

529 |
-> Int -- ^ Minimum delta parameter |

530 |
-> Int -- ^ Maximum delta parameter |

531 |
-> Maybe Solution -- ^ The best solution found (or Nothing) |

532 |
computeSolution nl bad_instances depth max_removals min_delta max_delta = |

533 |
let |

534 |
removals = computeRemovals nl bad_instances depth |

535 |
removals' = capRemovals removals max_removals |

536 |
in |

537 |
solutionFromRemovals removals' min_delta max_delta |

538 | |

539 |
-- Solution display functions (pure) |

540 | |

541 |
-- | Given the original and final nodes, computes the relocation description. |

542 |
computeMoves :: String -- ^ The instance name |

543 |
-> String -- ^ Original primary |

544 |
-> String -- ^ Original secondary |

545 |
-> String -- ^ New primary |

546 |
-> String -- ^ New secondary |

547 |
-> (String, [String]) |

548 |
-- ^ Tuple of moves and commands list; moves is containing |

549 |
-- either @/f/@ for failover or @/r:name/@ for replace |

550 |
-- secondary, while the command list holds gnt-instance |

551 |
-- commands (without that prefix), e.g \"@failover instance1@\" |

552 |
computeMoves i a b c d = |

553 |
if c == a then {- Same primary -} |

554 |
if d == b then {- Same sec??! -} |

555 |
("-", []) |

556 |
else {- Change of secondary -} |

557 |
(printf "r:%s" d, |

558 |
[printf "replace-disks -n %s %s" d i]) |

559 |
else |

560 |
if c == b then {- Failover and ... -} |

561 |
if d == a then {- that's all -} |

562 |
("f", [printf "migrate -f %s" i]) |

563 |
else |

564 |
(printf "f r:%s" d, |

565 |
[printf "migrate -f %s" i, |

566 |
printf "replace-disks -n %s %s" d i]) |

567 |
else |

568 |
if d == a then {- ... and keep primary as secondary -} |

569 |
(printf "r:%s f" c, |

570 |
[printf "replace-disks -n %s %s" c i, |

571 |
printf "migrate -f %s" i]) |

572 |
else |

573 |
if d == b then {- ... keep same secondary -} |

574 |
(printf "f r:%s f" c, |

575 |
[printf "migrate -f %s" i, |

576 |
printf "replace-disks -n %s %s" c i, |

577 |
printf "migrate -f %s" i]) |

578 | |

579 |
else {- Nothing in common -} |

580 |
(printf "r:%s f r:%s" c d, |

581 |
[printf "replace-disks -n %s %s" c i, |

582 |
printf "migrate -f %s" i, |

583 |
printf "replace-disks -n %s %s" d i]) |

584 | |

585 |
{-| Converts a placement to string format -} |

586 |
printSolutionLine :: InstanceList |

587 |
-> NameList |

588 |
-> NameList |

589 |
-> Int |

590 |
-> Int |

591 |
-> Placement |

592 |
-> Int |

593 |
-> (String, [String]) |

594 |
printSolutionLine il ktn kti nmlen imlen plc pos = |

595 |
let |

596 |
pmlen = (2*nmlen + 1) |

597 |
(i, p, s, c) = plc |

598 |
inst = Container.find i il |

599 |
inam = fromJust $ lookup (Instance.idx inst) kti |

600 |
npri = fromJust $ lookup p ktn |

601 |
nsec = fromJust $ lookup s ktn |

602 |
opri = fromJust $ lookup (Instance.pnode inst) ktn |

603 |
osec = fromJust $ lookup (Instance.snode inst) ktn |

604 |
(moves, cmds) = computeMoves inam opri osec npri nsec |

605 |
ostr = (printf "%s:%s" opri osec)::String |

606 |
nstr = (printf "%s:%s" npri nsec)::String |

607 |
in |

608 |
(printf " %3d. %-*s %-*s => %-*s %.8f a=%s" |

609 |
pos imlen inam pmlen ostr |

610 |
pmlen nstr c moves, |

611 |
cmds) |

612 | |

613 |
formatCmds :: [[String]] -> String |

614 |
formatCmds cmd_strs = |

615 |
unlines $ |

616 |
concat $ map (\(a, b) -> |

617 |
(printf "echo step %d" (a::Int)): |

618 |
(printf "check"): |

619 |
(map ("gnt-instance " ++) b)) $ |

620 |
zip [1..] cmd_strs |

621 | |

622 |
{-| Converts a solution to string format -} |

623 |
printSolution :: InstanceList |

624 |
-> NameList |

625 |
-> NameList |

626 |
-> [Placement] |

627 |
-> ([String], [[String]]) |

628 |
printSolution il ktn kti sol = |

629 |
let |

630 |
mlen_fn = maximum . (map length) . snd . unzip |

631 |
imlen = mlen_fn kti |

632 |
nmlen = mlen_fn ktn |

633 |
in |

634 |
unzip $ map (uncurry $ printSolutionLine il ktn kti nmlen imlen) $ |

635 |
zip sol [1..] |

636 | |

637 |
-- | Print the node list. |

638 |
printNodes :: NameList -> NodeList -> String |

639 |
printNodes ktn nl = |

640 |
let snl = sortBy (compare `on` Node.idx) (Container.elems nl) |

641 |
snl' = map (\ n -> ((fromJust $ lookup (Node.idx n) ktn), n)) snl |

642 |
m_name = maximum . (map length) . fst . unzip $ snl' |

643 |
helper = Node.list m_name |

644 |
header = printf |

645 |
"%2s %-*s %5s %5s %5s %5s %5s %5s %5s %5s %3s %3s %7s %7s" |

646 |
" F" m_name "Name" |

647 |
"t_mem" "n_mem" "i_mem" "x_mem" "f_mem" "r_mem" |

648 |
"t_dsk" "f_dsk" |

649 |
"pri" "sec" "p_fmem" "p_fdsk" |

650 |
in unlines $ (header:map (uncurry helper) snl') |

651 | |

652 |
-- | Compute the mem and disk covariance. |

653 |
compDetailedCV :: NodeList -> (Double, Double, Double, Double, Double) |

654 |
compDetailedCV nl = |

655 |
let |

656 |
all_nodes = Container.elems nl |

657 |
(offline, nodes) = partition Node.offline all_nodes |

658 |
mem_l = map Node.p_mem nodes |

659 |
dsk_l = map Node.p_dsk nodes |

660 |
mem_cv = varianceCoeff mem_l |

661 |
dsk_cv = varianceCoeff dsk_l |

662 |
n1_l = length $ filter Node.failN1 nodes |

663 |
n1_score = (fromIntegral n1_l) / (fromIntegral $ length nodes) |

664 |
res_l = map Node.p_rem nodes |

665 |
res_cv = varianceCoeff res_l |

666 |
offline_inst = sum . map (\n -> (length . Node.plist $ n) + |

667 |
(length . Node.slist $ n)) $ offline |

668 |
online_inst = sum . map (\n -> (length . Node.plist $ n) + |

669 |
(length . Node.slist $ n)) $ nodes |

670 |
off_score = (fromIntegral offline_inst) / |

671 |
(fromIntegral $ online_inst + offline_inst) |

672 |
in (mem_cv, dsk_cv, n1_score, res_cv, off_score) |

673 | |

674 |
-- | Compute the 'total' variance. |

675 |
compCV :: NodeList -> Double |

676 |
compCV nl = |

677 |
let (mem_cv, dsk_cv, n1_score, res_cv, off_score) = compDetailedCV nl |

678 |
in mem_cv + dsk_cv + n1_score + res_cv + off_score |

679 | |

680 |
printStats :: NodeList -> String |

681 |
printStats nl = |

682 |
let (mem_cv, dsk_cv, n1_score, res_cv, off_score) = compDetailedCV nl |

683 |
in printf "f_mem=%.8f, r_mem=%.8f, f_dsk=%.8f, n1=%.3f, uf=%.3f" |

684 |
mem_cv res_cv dsk_cv n1_score off_score |

685 | |

686 |
-- Balancing functions |

687 | |

688 |
-- Loading functions |

689 | |

690 |
{- | Convert newline and delimiter-separated text. |

691 | |

692 |
This function converts a text in tabular format as generated by |

693 |
@gnt-instance list@ and @gnt-node list@ to a list of objects using a |

694 |
supplied conversion function. |

695 | |

696 |
-} |

697 |
loadTabular :: String -> ([String] -> (String, a)) |

698 |
-> (a -> Int -> a) -> ([(String, Int)], [(Int, a)]) |

699 |
loadTabular text_data convert_fn set_fn = |

700 |
let lines_data = lines text_data |

701 |
rows = map (sepSplit '|') lines_data |

702 |
kerows = (map convert_fn rows) |

703 |
idxrows = map (\ (idx, (k, v)) -> ((k, idx), (idx, set_fn v idx))) |

704 |
(zip [0..] kerows) |

705 |
in unzip idxrows |

706 | |

707 |
-- | For each instance, add its index to its primary and secondary nodes |

708 |
fixNodes :: [(Int, Node.Node)] |

709 |
-> [(Int, Instance.Instance)] |

710 |
-> [(Int, Node.Node)] |

711 |
fixNodes nl il = |

712 |
foldl' (\accu (idx, inst) -> |

713 |
let |

714 |
assocEqual = (\ (i, _) (j, _) -> i == j) |

715 |
pdx = Instance.pnode inst |

716 |
sdx = Instance.snode inst |

717 |
pold = fromJust $ lookup pdx accu |

718 |
sold = fromJust $ lookup sdx accu |

719 |
pnew = Node.setPri pold idx |

720 |
snew = Node.setSec sold idx |

721 |
ac1 = deleteBy assocEqual (pdx, pold) accu |

722 |
ac2 = deleteBy assocEqual (sdx, sold) ac1 |

723 |
ac3 = (pdx, pnew):(sdx, snew):ac2 |

724 |
in ac3) nl il |

725 | |

726 |
-- | Compute the longest common suffix of a NameList list that |

727 |
-- | starts with a dot |

728 |
longestDomain :: NameList -> String |

729 |
longestDomain [] = "" |

730 |
longestDomain ((_,x):xs) = |

731 |
let |

732 |
onlyStrings = snd $ unzip xs |

733 |
in |

734 |
foldr (\ suffix accu -> if all (isSuffixOf suffix) onlyStrings |

735 |
then suffix |

736 |
else accu) |

737 |
"" $ filter (isPrefixOf ".") (tails x) |

738 | |

739 |
-- | Remove tails from the (Int, String) lists |

740 |
stripSuffix :: String -> NameList -> NameList |

741 |
stripSuffix suffix lst = |

742 |
let sflen = length suffix in |

743 |
map (\ (key, name) -> (key, take ((length name) - sflen) name)) lst |

744 | |

745 |
{-| Initializer function that loads the data from a node and list file |

746 |
and massages it into the correct format. -} |

747 |
loadData :: String -- ^ Node data in text format |

748 |
-> String -- ^ Instance data in text format |

749 |
-> (Container.Container Node.Node, |

750 |
Container.Container Instance.Instance, |

751 |
String, NameList, NameList) |

752 |
loadData ndata idata = |

753 |
let |

754 |
{- node file: name t_mem n_mem f_mem t_disk f_disk -} |

755 |
(ktn, nl) = loadTabular ndata |

756 |
(\ (name:tm:nm:fm:td:fd:fo:[]) -> |

757 |
(name, |

758 |
if any (== "?") [tm,nm,fm,td,fd] || fo == "Y" then |

759 |
Node.create 0 0 0 0 0 True |

760 |
else |

761 |
Node.create (read tm) (read nm) (read fm) |

762 |
(read td) (read fd) False |

763 |
)) |

764 |
Node.setIdx |

765 |
{- instance file: name mem disk status pnode snode -} |

766 |
(kti, il) = loadTabular idata |

767 |
(\ (name:mem:dsk:status:pnode:snode:[]) -> |

768 |
(name, |

769 |
Instance.create (read mem) (read dsk) |

770 |
status |

771 |
(fromJust $ lookup pnode ktn) |

772 |
(fromJust $ lookup snode ktn))) |

773 |
Instance.setIdx |

774 |
nl2 = fixNodes nl il |

775 |
il3 = Container.fromAssocList il |

776 |
nl3 = Container.fromAssocList |

777 |
(map (\ (k, v) -> (k, Node.buildPeers v il3 (length nl2))) nl2) |

778 |
xtn = swapPairs ktn |

779 |
xti = swapPairs kti |

780 |
common_suffix = longestDomain (xti ++ xtn) |

781 |
stn = stripSuffix common_suffix xtn |

782 |
sti = stripSuffix common_suffix xti |

783 |
in |

784 |
(nl3, il3, common_suffix, stn, sti) |

785 | |

786 |
-- | Compute the amount of memory used by primary instances on a node. |

787 |
nodeImem :: Node.Node -> InstanceList -> Int |

788 |
nodeImem node il = |

789 |
let rfind = flip Container.find $ il |

790 |
in sum . map Instance.mem . |

791 |
map rfind $ Node.plist node |

792 | |

793 |
-- | Compute the amount of disk used by instances on a node (either primary |

794 |
-- or secondary). |

795 |
nodeIdsk :: Node.Node -> InstanceList -> Int |

796 |
nodeIdsk node il = |

797 |
let rfind = flip Container.find $ il |

798 |
in sum . map Instance.dsk . |

799 |
map rfind $ (Node.plist node) ++ (Node.slist node) |

800 | |

801 | |

802 |
-- | Check cluster data for consistency |

803 |
checkData :: NodeList -> InstanceList -> NameList -> NameList |

804 |
-> ([String], NodeList) |

805 |
checkData nl il ktn _ = |

806 |
Container.mapAccum |

807 |
(\ msgs node -> |

808 |
let nname = fromJust $ lookup (Node.idx node) ktn |

809 |
nilst = map (flip Container.find $ il) (Node.plist node) |

810 |
dilst = filter (not . Instance.running) nilst |

811 |
adj_mem = sum . map Instance.mem $ dilst |

812 |
delta_mem = (truncate $ Node.t_mem node) |

813 |
- (Node.n_mem node) |

814 |
- (Node.f_mem node) |

815 |
- (nodeImem node il) |

816 |
+ adj_mem |

817 |
delta_dsk = (truncate $ Node.t_dsk node) |

818 |
- (Node.f_dsk node) |

819 |
- (nodeIdsk node il) |

820 |
newn = Node.setFmem (Node.setXmem node delta_mem) |

821 |
(Node.f_mem node - adj_mem) |

822 |
umsg1 = if delta_mem > 512 || delta_dsk > 1024 |

823 |
then [printf "node %s is missing %d MB ram \ |

824 |
\and %d GB disk" |

825 |
nname delta_mem (delta_dsk `div` 1024)] |

826 |
else [] |

827 |
in (msgs ++ umsg1, newn) |

828 |
) [] nl |