Viewstamped Replication Revisited
DSpace@MIT (Massachusetts Institute of Technology), 2012 · paper · Computer science
Presents an updated Viewstamped Replication, a technique for replicating services despite crashing nodes. It covers handling client requests, reorganizing after a replica fails, rejoining, optimizations, and a reconfiguration protocol that can change group membership and size.
Checked: PDF from cs.princeton.edu, fetched 4 Oct 2026. Page numbers are those of that PDF.
2 findings
Reference or label Section 4.1: "allow backups to process PREPARE messages out of order in Step 3" points to the wrong step; the backups' in-order processing rule is Step 4 ("Backups process PREPARE messages in order").
- p. 5 Section 4.1, paragraph after the normal-case steps
process PREPARE messages out of order in Step 3
- p. 4 Section 4.1, step 4
Backups process PREPARE messages in order
Judge's note: The page 4 crop shows that Step 3 ends with the primary sending ⟨PREPARE v, m, n, k⟩ to the other replicas. Step 4 begins "Backups process PREPARE messages in order: a backup won't accept a prepare with op-number n until...". The page 5 sentence that refers to backup processing cites Step 3, so the pointer should be Step 4.
Second check: another model agreed
Report a problem with this finding
Reference or label Reference [6] reads "ACM Transactions on Computer Systems 10, 2 (May 1998)"; the TOCS publication of The Part-Time Parliament is volume 16, issue 2.
- p. 14 References, entry [6]
Computer Systems 10, 2 (May 1998)
- p. 14 References [6]
ACM Transactions on Computer Systems 10, 2 (May 1998)
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that 'The Part-Time Parliament' appeared in ACM TOCS vol. 16, no. 2 (May 1998), pp. 133-169. The page 14 image shows "10, 2 (May 1998)". Internal support: ref [2] gives TOCS volume 20 for Nov. 2002, which fits volume 16 in 1998 (one volume per year), not volume 10. The page range is also missing.
Second check: another model agreed
Report a problem with this finding
11 minor (typos and the like; most checked by the judge only)
Reference or label Reference [1] dates Castro's thesis "MIT-LCS-TR-817 ... Cambridge, Jan. 2000. Ph.D. thesis."; the technical report is generally dated January 2001.
- p. 14 References, entry [1]
Cambridge, Jan. 2000. Ph.D. thesis.
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that MIT-LCS-TR-817 (Castro, Practical Byzantine Fault Tolerance) is dated January 2001. The page 14 image shows "Jan. 2000". This cannot be checked within the files, so the authors should confirm the date.
Second check: another model agreed
Report a problem with this finding
Wording View-change step 5 says twice "and update the information in their client-table" (once after setting status to normal, and again after executing committed operations), so it is unclear when the update happens.
- p. 6 Section 4.2, view change step 5
and update the information in their client-table
Judge's note: The page 6 crop of step 5 shows the identical clause "and update the information in their client-table" twice within the same step, before and after "Then they execute all operations known to be committed...". The text does not say whether these are two different updates. The redundancy leaves the order ambiguous.
Second check: another model agreed
Report a problem with this finding
Wording The Conclusions say the paper "describes ways to reduce latency of reads and writes", but the only latency optimization in the body is for reads (Section 6.3). Batching is said to have "little impact on latency" (Section 6.2).
- p. 14 Section 9, Conclusions
ways to reduce latency of reads and writes
- p. 9 Sections 6.2–6.3
This way it has little impact on latency.
Only the authors can say what is right here.
Judge's note: I searched the text for latency and delay. Section 6.3 opens with "two ways to reduce the latency for handling of read requests". Section 6.2 says batching "has little impact on latency". The delay discussion in Section 7 concerns reconfiguration, not write latency. I found no technique presented as reducing write latency. Only the authors can say whether something such as disk-free processing was meant.
Second check: another model agreed
Report a problem with this finding
Typo Section 2.2: "there is a least one replica that knows about the request" should read "at least one replica".
- p. 2 Section 2.2, second paragraph
there is a least one replica that knows about the request
- p. 2 Section 2.2, second paragraph
there is a least one replica
- p. 2 Section 2.2, rationale paragraph
there is a least one replica that
Judge's note: The page 2 crop shows "there is a least one replica that" in the left column. This is a word-level typo and not an extraction artefact.
Second check: none (judge only)
Report a problem with this finding
Typo Section 4.2 step 1: "message to the all other replicas" has a stray article; it should be "to all other replicas".
- p. 5 Section 4.2, view change step 1
message to the all other replicas
Judge's note: The page 5 crop of view-change step 1 shows "message to the all other replicas, where v identifies the new view".
Second check: none (judge only)
Report a problem with this finding
Typo Inside the message tuples the primed fields are set with a text apostrophe ("⟨DOVIEWCHANGE v, l, v’, n, k, i⟩", "⟨GETSTATE v, n’, i⟩"), but the explanatory prose uses a math prime (v′, n′).
- p. 5 Section 4.2, view change step 2
DOVIEWCHANGE v, l, v’, n, k, i
- p. 8 Section 5.2
GETSTATE v, n’,
Judge's note: The crops of pages 5 and 8 show a curly apostrophe after the italic v and n in the tuples, and a math prime in the definitions that follow ("v′ is the view number...", "n′ is its op-number"). The meaning is clear, so this is a cosmetic typographic inconsistency.
Second check: none (judge only)
Report a problem with this finding
Typo Section 4.2: "taking the log for the next view from latest previous active view and ignoring logs from earlier view" should read "from the latest previous active view ... from earlier views".
- p. 6 Section 4.2, paragraph after step 5
ignoring logs from earlier view
Judge's note: The page 6 crop shows the sentence as quoted. This is a grammatical slip only.
Second check: none (judge only)
Report a problem with this finding
Typo Section 6.3.1: "assuming clocks rates are loosely synchronized" should read "clock rates".
- p. 9 Section 6.3.1, leases paragraph
assuming clocks rates are loosely synchronized
Judge's note: The page 9 crop of the leases paragraph shows "assuming clocks rates are loosely synchronized".
Second check: none (judge only)
Report a problem with this finding
Typo Sections 7.1 and 7.3: the possessive is typeset in math mode as a prime, "s is c′s request-number", which reads as a variable c′ rather than "c's".
- p. 10 Section 7.1, RECONFIGURATION message description
s is c′s request-number
- p. 11 Section 7.3, CHECKEPOCH description
s is c′s request-number
Judge's note: The crops of pages 10 and 11 both show c with a math prime (c′s) in the RECONFIGURATION and CHECKEPOCH descriptions. Elsewhere the paper uses primes for distinct quantities (v′, n′, f′), so the typesetting is misleading. The fix is evident.
Second check: none (judge only)
Report a problem with this finding
Typo Section 8.2 Liveness: "assuming no more that f replicas fail simultaneously" should read "no more than f".
- p. 13 Section 8.2, Liveness
assuming no more that f replicas fail simultaneously
- p. 13 Section 8.2, Liveness
assuming no more that f replicas
Judge's note: The page 13 crop shows "assuming no more that f replicas". This is a word-level typo.
Second check: none (judge only)
Report a problem with this finding
Typo Section 8.2: "before sending the DoViewChange messages" sets the message name in italic mixed case, whereas everywhere else it is in small caps (DOVIEWCHANGE), including in the same section.
- p. 13 Section 8.2, paragraph before Liveness
before sending the DoViewChange messages
- p. 13 Section 8.2, paragraph before Liveness
DoViewChange messages
Judge's note: The page 13 crop shows italic "DoViewChange" in the right column. A few lines earlier the left column has small-caps DOVIEWCHANGE. This is a cosmetic notation inconsistency.
Second check: none (judge only)